<%BANNER%>

Proton Radiation and Thermal Stability of Gallium Nitride and Gallium Nitride Devices


PAGE 1

PROTON RADIATION AND THERMAL STABILTY OF GALLIUM NITRIDE AND GALLIUM NITRIDE DEVICES By KIMBERLY K. ALLUMS 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 Kimberly K. Allums

PAGE 3

This document is dedicated to my Family and Friends. Thank you for your continued support of my endeavors and most of all me.

PAGE 4

iv ACKNOWLEDGMENTS First and foremost I’d like to thank God, because without Him I wouldn’t be where I am today. I’d like to thank my family for being my support throughout my entire life and always pushing me to do my best. I’d also like to thank Dr. Cammy R. Ab ernathy for accepting me into her group and for her guidance through my graduate curr iculum. I thank my committee members for their guidance and suggestions. I especially thank Dr. Brent Gila for all his tutelage during my research work. I thank Abernathy Electronics Group for their friendship and help with my research. I’d like to thank Dr. Richard Wilkins, Mr. Shoj a Ardalon and Mr. Dwevdi of Prairie View A&M University for collaborating with us on th is project. I appreciat e all that every one has done. This work was supported by ONR Grant No. US Navy N00014-98-1-0204, and the Prairie View A&M University work wa s supported by NASA Grant Nos. NCC 9-114.

PAGE 5

v TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES...........................................................................................................viii LIST OF FIGURES...........................................................................................................ix ABSTRACT.....................................................................................................................xv i CHAPTER 1 INTRODUCTION........................................................................................................1 2 BACKGROUND AND LI TERATURE REVIEW......................................................3 2.1 Introduction.............................................................................................................3 2.2 Device Structures....................................................................................................3 2.2.1 Metal Oxide Semiconductor Field Effect Transistor..................................3 2.2.2 Metal Oxide Semiconductor Capacitors.......................................................5 2.3 Growth and Photoluminescence Spectroscopy of Gallium Nitride........................6 2.4 Dielectrics.............................................................................................................10 2.4.1 Ideal Dielectrics..........................................................................................10 2.4.2 Crystalline vs. Amorphous.........................................................................11 2.4.3 Present State of Dielectrics for Gallium Nitride.........................................12 2.4.3.1 Silicon oxide on GaN.......................................................................13 2.4.3.2 Silicon nitride on GaN......................................................................14 2.4.3.3 Aluminum nitride on GaN................................................................14 2.4.3.4 Gallium oxide on GaN.....................................................................15 2.4.3.5 Silicon dioxide on gallium oxide on GaN........................................16 2.4.3.6 Gallium gadolinium oxide on GaN..................................................17 2.4.3.7 Gadolinium oxide on GaN...............................................................17 2.4.3.8 Scandium oxide on GaN..................................................................17 2.5 Epitaxal Growth....................................................................................................18 2.5.1 Molecular Beam Epitaxy............................................................................19 2.5.2 Substrate Preparation..................................................................................19 2.6 Types and Effects of Radi ation on Microelectronics...........................................21 2.7 Previous Radiation Studies...................................................................................23

PAGE 6

vi 3 EXPERIMENTAL APPROACH...............................................................................35 3.1 Alternative Dielectrics..........................................................................................35 3.2 Oxide Growth Parameters.....................................................................................36 3.2.1 Scandium Oxide Growth............................................................................36 3.2.2 Magnesium Oxide Growth.........................................................................36 3.2.3 Magnesium Calcium Oxide Growth...........................................................36 3.3 Materials Characterization....................................................................................37 3.3.1 Auger Electron Spectroscopy.....................................................................37 3.3.2 Atomic Force Microscopy..........................................................................37 3.3.3 Transmission Electron Microscopy............................................................38 3.3.4 Scanning Electron Microscopy...................................................................38 3.3.5 X-Ray Reflectivity......................................................................................39 3.3.6 Photoluminescence.....................................................................................39 3.3.7 Hall Effect..................................................................................................40 3.3.8 Current Voltage Analysis.........................................................................40 3.1.9 Capacitance – Voltage Analysis.................................................................41 3.2 Diode Fabrication.................................................................................................41 3.3 Proton Radiation Setup and Facility.....................................................................42 4 RADIATION AND THERMAL STABILITY OF VARIOUS TYPES OF GALLIUM NITRIDE.................................................................................................53 4.1 Characterization of MOCVD N-type GaN...........................................................53 4.2 Characterization of MOCVD U-type GaN...........................................................57 4.3 Characterization of MB E Grown Ga-Polar GaN..................................................58 4.4 Characterization of MOCVD P-type GaN............................................................58 4.5 Summary...............................................................................................................59 5 RADIATION AND THERMAL STAB ILITY OF NOVEL OXIDES ON GALLIUM NITRIDE...............................................................................................106 5.1 Characterization of Sc2O3 on GaN.....................................................................107 5.2 Characterization of Sc2O3/MgO on GaN............................................................108 5.3 Characterization of Sc2O3/MgCaO on GaN.......................................................109 5.4 Summary.............................................................................................................109 6 PROCESSING AND THERMAL ST ABILITY OF MOS DEVICES.....................141 6.1 Processing of Novel Oxid e/Gallium Nitride Devices.........................................141 6.2 Proton Radiation Effects of GaN MOS Diodes..................................................142 6.3 Effects of In-situ and Ex-situ Thermal Annealing..............................................143 6.4 Summary.............................................................................................................146 7 SUMMARY AND DISSCUSSION.........................................................................172 7.1 GaN vs. Oxide/GaN............................................................................................172 7.2 MOS Devices......................................................................................................173

PAGE 7

vii LIST OF REFERENCES.................................................................................................176 BIOGRAPHICAL SKETCH...........................................................................................181

PAGE 8

viii LIST OF TABLES Table page 2-1 Properties of Dielectrics Previ ously Studied for Use with GaN..............................28 3-1 Proposed Oxide Properties.......................................................................................43 6-1 Oxide Etchants and Condition................................................................................147 6-2 Dit Values at .4eV for Oxide/GaN Devices...........................................................148

PAGE 9

ix LIST OF FIGURES Figure page 2-1 Typical MOSFET...................................................................................................29 2-2 Cross-section illustration of a enhancement mode MOSFET................................30 2-3 Cross section of a MOS capacitor..........................................................................31 2-4 Dipole formation in the pres ence of an electric field.............................................31 2-5 Sketch of Riber MBE used fo r oxide growth (after Gila)......................................32 2-6 AFM images of as received GaN. A) MOCVD GaN from Epitronics B) as received MBE GaN from SVT..............................................................................33 2-7 Examples of GaN surfaces before gr owth. A) The UV-ozone treated surface of GaN. B) Buffered oxide etched surface of GaN....................................................34 2-8 Photos of RHEED indicating a (1x3) pa ttern. A) <11-20> cr ystal direction. B) <1-100> crystal direction.......................................................................................34 3-1 Auger Electron Spectroscopy set-up......................................................................44 3-2 Atomic force microscope (after K.K. Harris 2000)...............................................45 3-3 TEM setup used to image atomic la yers at the film/substrate interface................46 3-4 SEM operation. Electron beam is rastered over the sample producing secondary electrons (aft er S.M. Donovan 1999)...................................................47 3-5 Photoluminescence setup.......................................................................................48 3-6 Ideal MOS I-V curve.............................................................................................49 3-7 Ideal C-V curves of n-type MOS (after Johnson)..................................................49 3-8 Finished MOS capacitors design...........................................................................50 3-9 Test Chamber for Radiation. A) Mounted samples in test chamber. B) Shutters of beam...................................................................................................................51

PAGE 10

x 3-10 Radiation Test Facility at Texas A&M University................................................52 4-1 Initial PL scans of non-radiated 1s t UOE n-GaN at temperatures 300K and 15K.........................................................................................................................61 4-2 Room Temperature PL analysis of 1s t UOE n-GaN before and after radiation....62 4-3 15K PL analysis of 1st UOE n-GaN before and after radiation............................63 4-4 Comparison of 300K 1st UOE n-GaN PL scans, pre-radiation, immediately post radiation, and post radiation 1 year later........................................................64 4-5 PL analysis of annealed un-radiat ed 1st UOE n-GaN. A) PL Spectra B) Bandedge inset.......................................................................................................65 4-6 PL analysis of annealed 10MeV radi ated 1st UOE n-GaN A) PL Spectra B) Bandedge inset (below)..........................................................................................66 4-7 PL analysis of annealed 40MeV radi ated 1st UOE n-GaN. A) PL Spectra B) Bandedge inset.......................................................................................................67 4-8 Normalized bandedge peaks of all annealed 1st UOE GaN to the 1st UOE GaN control peak intensity of 1.4812 vs. the annealing temperatures. 0 temperature indicates the initial state of the GaN......................................................................68 4-9 Pre and post radiation PL analysis of 2nd UOE n-GaN..........................................69 4-10 PL analysis of un-radiated anneal ed 2nd UOE n-GaN. A) PL Spectra B) Bandedge inset.......................................................................................................70 4-11 PL analysis of annealed 10MeV ra diated UOE n-GaN. A) PL Spectra B) Bandedge inset.......................................................................................................71 4-12 PL analysis of annealed 40MeV ra diated UOE n-GaN. A) PL Spectra B) Bandedge inset.......................................................................................................72 4-13 Normalized bandedge peaks of annealed 2nd UOE n-GaN to the 2nd UOE nGaN control peak intensity of 12.354 vs. the annealing temperatures..................73 4-14 Pre and post radiation PL analysis of UF n-GaN...................................................74 4-15 PL analysis of annealed un-radiated UF n-GaN. A) PL Spectra B) Bandedge inset........................................................................................................................75 4-16 PL analysis of annealed 10MeV ra diated UF n-GaN. A) PL Spectra B) Bandedge inset.......................................................................................................76 4-17 PL analysis of annealed 40MeV radi ated UF n-GaN. A) PL Spectra B) Bandedge inset.......................................................................................................77

PAGE 11

xi 4-18 Normalized bandedge peaks of anneal ed UF n-GaN to the UF n-GaN control peak intensity of 25.069 vs. the annealing temperatures.......................................78 4-19 Pre and Post Radi ation of UOE u-GaN..................................................................79 4-20 PL analysis of annealed un-radiated UOE u-GaN. A) PL Spectra B) Bandedge inset........................................................................................................................80 4-21 PL analysis of annealed 10MeV ra diated UOE u-GaN. A) PL Spectra B) Bandedge inset.......................................................................................................81 4-22 PL analysis of annealed 40MeV ra diated UOE u-GaN. A) PL spectra B) Bandedge inset.......................................................................................................82 4-23 Normalized bandedge peaks of a nnealed UOE u-GaN to the UOE u-GaN control peak intensity of 0.1162 vs. the annealing temperatures...........................83 4-24 Pre and post radiation PL analysis of UF u-GaN...................................................84 4-25 PL analysis of annealed un-radiated UF u-GaN. A) PL Spectra B) Bandedge inset........................................................................................................................85 4-26 PL analysis of annealed 10MeV ra diated UF u-GaN. A) PL Spectra B) Bandedge inset.......................................................................................................86 4-27 PL analysis of annealed 40MeV ra diated UF u-GaN. A) PL Spectra B) Bandedge inset.......................................................................................................87 4-28 Normalized bandedge peaks of anneal ed UF u-GaN to the UF u-GaN control peak intensity of 1.2997 vs. the annealing temperatures.......................................88 4-29 Pre and Post Radiation PL an alysis of MBE Ga-polar GaN..................................89 4-30 PL analysis of annealed un-radiated MBE Ga-polar GaN. A) PL Spectra B) Bandedge inset.......................................................................................................90 4-31 PL analysis of annealed 10MeV ra diated MBE GaN. A) PL Spectra B) Bandedge inset.......................................................................................................91 4-32 PL analysis of annealed 40MeV ra diated MBE GaN. A) PL Spectra B) Bandedge inset.......................................................................................................92 4-33 Normalized bandedge peaks of anneal ed MBE Ga-polar GaN to the MBE Gapolar GaN control peak intensity of .01016 vs. the annealing temperatures.........93 4-34 Comparison of PL scans of as grown P-GaN and Activated P-GaN.....................94 4-35 Pre and post radiation PL analysis of as-grown P-GaN.........................................95

PAGE 12

xii 4-36 Pre and post radiation PL an alysis of Activated P-GaN........................................96 4-37 PL analysis of annealed as-grown un-radiated P-GaN..........................................97 4-38 PL analysis of annealed un-radiated Activated P-GaN..........................................98 4-39 PL analysis of anneal ed 10MeV as-grown P-GaN................................................99 4-40 PL analysis of anneal ed 40MeV as-grown P-GaN..............................................100 4-41 PL analysis of annealed Activ ated P-GaN radiated at 10MeV............................101 4-42 PL analysis of annealed Activ ated P-GaN radiated at 40MeV............................102 4-43 Normalized bandedge peaks of anneal ed as grown P-GaN to the as grown PGaN control peak intensity of 2.7808 vs. the annealing temperatures................103 4-44 Normalized bandedge peaks of anneal ed Activated P-GaN to the Activated PGaN control peak intensity of .42224 vs. the annealing temperatures................104 4-45 All Normalized P-GaN bandedge peaks to control peaks of 2.7808 and .42224 respectively..........................................................................................................105 5-1 Preliminary Sc2O3/GaN A) PL spectra and B) XRR scan...................................111 5-2 Sc2O3/GaN PL analysis. A) PL at 300K. B) PL at 15K.......................................112 5-3 Sc2O3/GaN XRR scans showing the change in oscillations................................113 5-4 Sc2O3/GaN PL scans of pre-radiation, immediate post radiation and post irradiation 1 year later. A) PL spectra B) Bandedge inset...................................114 5-5 PL scans of samples annealed non-radiated Sc2O3/GaN to 900C. A) PL Spectra B) Bandedge inset................................................................................................115 5-6 Sc2O3/GaN post radiation anneals to 900oC for radiation energies of 10MeV. A) PL spectra B) Bandedge inset.........................................................................116 5-7 Sc2O3/GaN post radiation anneals to 900oC for radiation energies of 40MeV. A) PL Spectra B) Bandedge inset.......................................................................117 5-8 Sc2O3/GaN XRR scans of unradiated annealed sample at 400C.........................118 5-9 Sc2O3/GaN XRR scans pre and post radiation annealed at 400oC.......................119 5-10 Normalized bandedge peaks of annealed Sc2O3/GaN to the Sc2O3/GaN control peak intensity of .5139 vs. annealing temperature...............................................120

PAGE 13

xiii 5-11 Preliminary Sc2O3(20nm)/MgO(20nm)/GaN PL spectra and XRR. A) PL spectra B) XRR scan............................................................................................121 5-12 Sc2O3(20nm)/MgO(20nm)/GaN PL scans. A) PL at 300K B) PL at 15K..........122 5-13 Sc2O3(20nm)/MgO(20nm)/GaN XRR scan showing the change in oscillations at 40 MeV............................................................................................................123 5-14 Sc2O3(20nm)/MgO(20nm)/GaN PL scans of pre-radiation, immediate post radiation and post irradiati on 1 year later. A) PL spectra B) Bandedge inset.....124 5-15 PL scans of samples annealed non-radiated Sc2O3(20nm)/MgO(20nm)/GaN to 900C. A) Pl Spectra B) Bandedge inset...............................................................125 5-16 Sc2O3(20nm)/MgO(20nm)/GaN post radiation anneals to 900oC for radiation energies of 10MeV. A) PL spectra B) Bandedge inset........................................126 5-17 Sc2O3(20nm)/MgO(20nm)/GaN post radiation anneals to 900oC for radiation energies of 40MeV. A) PL spectra B) Bandedge inset........................................127 5-18 Sc2O3(20nm)/MgO(20nm)/GaN XRR scan of annealed sample........................128 5-19 Sc2O3(20nm)/MgO(20nm)/GaN XRR scans pre and post radiation annealed at 400oC....................................................................................................................129 5-20 Normalized bandedge peaks of annealed Sc2O3(20nm)/MgO(20nm)/GaN to the Sc2O3(20nm)/MgO(20nm)/GaN contro l peak intensity of .21556 vs. annealing temperature..........................................................................................130 5-21 Preliminary Sc2O3(20nm)/MgCaO(20nm)/GaN. A) PL spectra B) XRR scan...131 5-22 Sc2O3(20nm)/MgCaO(20nm)/GaN PL scans. A) PL at 300K B) PL at 15K (bottom)................................................................................................................132 5-23 Sc2O3(20nm)/MgCaO(20nm)/GaN XRR scan showing the change in oscillations...........................................................................................................133 5-24 Sc2O3(20nm)/MgCaO(20nm)/GaN PL scans of pre-radiation, immediate post radiation and post irradiati on 1 year later. A) PL spectra B) Bandedge inset.....134 5-25 PL scans of samples annealed non-radiated Sc2O3(20nm)/MgCaO(20nm)/GaN to 900C. A) Pl Spectra B) Bandedge inset...........................................................135 5-26 Sc2O3(20nm)/MgCaO(20nm)/GaN post radiation anneals to 900oC for radiation energies of 10MeV. A) PL spectra B) Bandedge inset.........................136 5-27 Sc2O3(20nm)/MgCaO(20nm)/GaN post radiation anneals to 900oC for radiation energies of 40MeV. A) PL spectra B) Bandedge inset.........................137

PAGE 14

xiv 5-28 Sc2O3(20nm)/MgCaO(20nm)/GaN XRR scan of annealed sample....................138 5-29 Sc2O3(20nm)/MgCaO(20nm)/GaN XRR scans pre and post radiation annealed at 400oC................................................................................................................139 5-30 Normalized bandedge peaks of annealed Sc2O3(20nm)/MgCaO(20nm)/GaN to the Sc2O3(20nm)/MgCaO(20nm)/GaN contro l peak intensity of 1.2968 vs. annealing temperature..........................................................................................140 6-1 Finished diode mask design.................................................................................149 6-2 Ohmic IV plots taken from sa mples receiving various treatments......................150 6-3 Current density –Voltage behavior of Sc2O3/GaN after 10MeV irradiation.......151 6-4 Capacitance –Voltage plot of Sc2O3/GaN after 10MeV irradiation....................152 6-5 Current density –Voltage behavior of Sc2O3/ MgO/GaN after 10MeV irradiation.............................................................................................................153 6-6 Capacitance –Voltage plot of Sc2O3/ MgO/GaN after 10MeV irradiation..........154 6-7 Comparison of Current Density -Voltage results of different Sc2O3/MgCaO/GaN thickness.............................................................................155 6-8 Capacitance-Voltage plot of Sc2O3/MgCaO/GaN after 10M eV irradiation........156 6-9 Capacitance-Voltage plot of Sc2O3/MgCaO/GaN after 10M eV irradiation........157 6-10 Dit of post 10MeV radiation. A) Sc2O3/GaN B) Sc2O3/MgO/GaN.....................158 6-11 Current Density –Vol tage behavior of Sc2O3/GaN during and after thermal annealing..............................................................................................................159 6-12 Capacitance –Voltage plot of Sc2O3/GaN during and after heating....................160 6-13 Current Density –Vol tage behavior of Sc2O3/MgO/GaN during and after heating..................................................................................................................161 6-14 Capacitance –Voltage breakdown of Sc2O3/MgO/GaN after heating.................162 6-15 Current Density –Vol tage behavior of Sc2O3/5nm MgCaO/GaN during and after heating.........................................................................................................163 6-16 Capacitance-Voltage plot of Sc2O3/MgCaO/GaN during and after heating........164 6-17 Capacitance-Voltage plot of Sc2O3/MgCaO/GaN during and after heating. A) CV of in-situ testing. B) CV of sa mples cooled to room temperature.................165

PAGE 15

xv 6-18 Capacitance –Voltage plot of Sc2O3/GaN after forming gas anneal at 400oC.....166 6-19 Capacitance –Voltage plot of Sc2O3/MgO/GaN after forming gas anneal at 400oC....................................................................................................................167 6-20 Capacitance-Voltage plot of Sc2O3/MgCaO/GaN after forming gas anneal at 400oC....................................................................................................................168 6-21 Dit of Annealed Sc2O3/GaN. A) Probe station thermal anneals. B) Forming Gas anneal............................................................................................................169 6-22 Dit of Annealed 2nm Sc2O3/40nm MgO/GaN. A) Probe station thermal anneals. B) Forming gas anneal...........................................................................170 6-23 Dit of Annealed 35nm Sc2O3/5 nm MgCaO/GaN A) Probe station thermal anneals. B) Forming gas anneal...........................................................................171 7-1 Recovery of Oxide/GaN post 900oC after 7 months............................................175

PAGE 16

xvi Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy PROTON RADIATION AND THERMAL STABILITY OF GALLIUM NITRIDE AND GALLIUM NITRIDE DEVICES By Kimberly K. Allums May 2006 Chair: Cammy R. Abernathy Major Department: Materials Science and Engineering In today’s industry one can see a constant challenge to exceed the limits of yesterday’s devices. For the last three d ecades, the III–V nitride semiconductors have been viewed as highly promising for semi conductor device applications. The primary focus of III–V nitrides, thus far, has b een centered on light emitting diodes (LEDs), injection lasers for digital data reading and storage applicatio ns, and ultra violet photodetectors. Yet, another application is hi gh-power electronic de vices for space-borne communications systems. It is expected that GaN-based devices will be more resistant to radiation damage often encountered in space environments, though verification of this is just now being undertaken. In particular, no information is yet available about the sensitivity to radiation of devices using di electrics such as MOSFETs. Similarly, very limited data has been reported on the e ffects of high-energy protons on GaN based devices of any type. For this reason the re search presented in this dissertation was undertaken to study the radiation and thermal stability of gallium nitride materials and

PAGE 17

xvii gallium nitride semiconductor diodes, with a nd without novel gate dielectrics such as, scandium oxide (Sc2O3) and magnesium oxide (MgO) and the ternary mix of magnesium calcium oxide (MgCaO). It was found that though environmental de gradation could be a problem for MgO dielectrics, the radiation expos ure itself did not produce signif icant damage in either the Sc2O3, MgO or MgCaO dielectrics. Much of the minimal damage occurred in the GaN as shown by photoluminescence spectroscopy (PL).

PAGE 18

1 CHAPTER 1 INTRODUCTION In today’s industry, microelectronics has b een largely based on silicon solid-state devices. As technology has improved, the demand for devices that can operate at higher temperatures and in more caustic environments has become the focus of several research areas especially military applications. As a result, compound semiconductors are becoming increasingly important because they do not possess the limitations of Sidevices and have direct bandgaps that allow them to be us ed in optical applications. Much of the focus of compound semiconductors has been on the III-V groups. Compound semiconductors, such as gallium arsenide (GaAs) and indium phosphide (InP), have higher carrier mobilities and lower-s aturation electric fields than silicon semiconductor devices, and research based on these compound semiconductors have led to breakthroughs in device performance. For th e last few decades the III-V nitrides, such as gallium nitride (GaN), have been vi ewed as highly promising for semiconductor device applications mainly because of a wi der bandgap that will allow the semiconductor to overcome thermal and power handling limits of the GaAs and InP semiconductors. The primary focus of GaN applications thus fa r has centered on light emitting diodes (LEDs), injection lasers for digital data storage and ultraviolet photode tectors. Because of its large bandgap, Eg, GaN can operate at higher temper atures than its other semiconductor counterparts. However, the large Eg also requires a large ba ndgap oxide in order to provide adequate carrier conf inement in MOS-type applications. Various oxides were tested in previous studies but have been proven to possess many limitations as gate

PAGE 19

2 dielectrics. Some alternative materials that are candidates for an optimum dielectric for GaN are magnesium oxide (MgO), and a te rnary combination of magnesium calcium oxide (MgCaO). For use in wide bandgap se miconductor devices, the dielectric materials must also possess excellent thermal stab ility, both because of the high operating temperatures and the high processing temper atures needed for device fabrication. These oxides must also be radiation resistant for use in low orbit and aerospace applications. The ultimate goal of this project is to provi de a high quality oxide that can improve the radiation hardness and therma l stability of GaN devices.

PAGE 20

3 CHAPTER 2 BACKGROUND AND LITERATURE REVIEW 2.1 Introduction This chapter discusses the fundamentals of field effect transistors (FET), metal oxide semiconductor capacitors, growth and char acterization of gallium nitride, dielectric films and their properties, oxide growth pro cesses and device applications. Also included is a literature review of the possible effects of radiat ion and the characterization methods used to determine the stability of the heterostructures in question. 2.2 Device Structures A field effect transistor (F ET) is a unipolar device wh ere only one type of carrier takes part in the conduction pr ocess. The FET is a three-te rminal device in which the current through two terminals is controlled by a voltage applied at the third. FETs are characterized by high input impedance since th e control voltage is applied to a reverse biased junction either through a metal Schottky barrier or across an insulator [1]. Some advantages of FETs include higher switching speeds and higher cutoff frequencies than bipolar devices. There are vari ous types of FETs such as j unction field eff ect transistors (JFET) and metal semiconductor field effect tr ansistors (MESFET), but the focus of this study will be metal oxide semiconductor fiel d effect transistors (MOSFET) and metal oxide semiconductor (MOS) capacitors. 2.2.1 Metal Oxide Semiconducto r Field Effect Transistor The MOSFET is one of the most important devices for very large-scale integrated circuits such as microprocessors, and semi conductor memories and is also becoming an

PAGE 21

4 important power device [2]. Similar devices to the MOSFET include the metal insulator semiconductor field effect tran sistor (MISFET) and the in sulated gate field effect transistor (IGFET). For all of these device types, the gate region of the transistor determines the capabilities of the device. The uniqueness of the MOSFET relies on the oxide layer’s ability to prevent current flow from the semiconductor to the gate due to the high resistivity of the oxide. Figure 2-1 show s a general structure of a MOSFET. Ideally a MOSFET should possess high output current drive, ID, high transconductance, gm, stable threshold voltage, VT, fast switching speed, high gate oxide breakdown voltage and low source/drain to body capacitanc e. Physically, the current drive is linearly related to gate width and as ID increases so does the capacitance. There are two types of MOSFET devices, depletion mode and enhancement mode. In the depletion mode device, the material under the gate is dope d in order to carry current. This device is in the “ON” state wh en there is no applied gate voltage. When there is a zero gate voltage, carriers are free to flow from the sour ce to the drain in the MOSFET structure. As a negative voltage is applied to the gate contact, the area under the gate, called the channel, is gradually depl eted of carriers. Th e depletion depth is increased until the flow from source to drain is stopped. This voltage is called the pinch off voltage, since it effectively pinches the chan nel shut. The transistor is now “OFF.” As the voltage across the sour ce-drain is increased, it requi res more gate voltage to successfully pinch-off the carrier flow. In the enhancement mode device, the material type under the gate is not doped; thus no cha nnel is present and in an OFF state since there is no current flow. As with the deplet ion mode MOSFET, a positive drain voltage is still applied but there is also a positive gate vo ltage applied as well. This has the effect of

PAGE 22

5 attracting free electrons towards the gate, thus creating a channel of free flowing electrons and the larger the positive gate voltage the wider the carrier channel of electrons. Figure 2-2 shows an example of the ON and OFF states of the enhancement mode device. Thus the maximum operating parameters of the device are determined by the amount of electric field that can be appl ied to the gate before dielectric breakdown occurs. Also, forward gate voltages can be us ed to increase the amount of current, which can be passed through the channel. Again a high dielectric breakdown field is required. The MOSFET has several advantages over hete rojunction type transistors, such as relative insensitivity to temperature dur ing operation. Also MOSFET devices are expected to have a wide gate modulation range. This wide range is because the device turn-on is dependent on the dielectric thickness and is not limited to low turn-on voltages like those obtained when usi ng Schottky metal contacts. 2.2.2 Metal Oxide Semiconductor Capacitors The MOS capacitor (MOSC) is the most basic device used to eval uate the electrical properties of the oxide and the semiconducto r. Throughout this study the MOS capacitor, also known as the MOS diode, is used for eval uation of the MOS structure. For use with GaN on sapphire substrate, the contacts are cr eated front side. The process is done by etching the oxide to expose the underlying semiconductor material and then depositing ohmic and gate contacts. Figure 2-3 shows a cross section of the MOSC. The MOSC is operated by applying a potential difference across the gate and ohmic contacts. An understanding of MOSC behavior is obtaine d through the use of energy band diagrams and ideal capacitance-voltage curves.

PAGE 23

6 2.3 Growth and Photoluminescence Spectroscopy of Gallium Nitride One of the ways high quality gallium nitrid e is grown is via me tal organic chemical vapor deposition (MOCVD). Trimethl gallium (TMG) and ammonia (NH3) are used as source gases for Group III and V species to obtai n c-axis oriented films of GaN grown on (0001) sapphire (Al2O3) wafers. In MOCVD the flow of th e reactant gas sources is very critical to the GaN film grow th and any slight change can alter the film quality. Hydrogen is normally used as a carrier gas and is flow ed normal to the substrate surface to bring the reactant gases in contact with the substrat e and to prevent thermal convection effects. MOCVD can be used to grow n-type and p-type GaN. Undop ed GaN is naturally n-type due to nitrogen vacancies that occur in the film. Two types of GaN are generally mentioned when referring to n-type GaN, doped and unintentially doped. Silicon is normally used to doped GaN to increase the number of electron ca rriers within the substrate and the carrier con centration is typically ~1019cm-3 whereas unintentially doped GaN (known as u-GaN) has a typical intrinsi c electron carrier con centration of ~1016 cm3. As well as being n-type, MOCVD GaN is nor mally grown with N terminated faces at the surface which tend to have dangling bonds that allow for the surface to be susceptible to contamination and impurities. This can lead to surface effects such as surface charging and fermi level pinning. Ptype GaN is made by doping the GaN with magnesium, Mg, to create majority carriers of holes. Gallium nitride can also be grown via mol ecular beam epitaxy (MBE) with either N terminated ([0001] direction) or Ga terminated ([000-1] direction) su rfaces using sapphire substrates. MBE’s inherent co ntrol over growth parameters can be used to interrogate certain structural and electrical processes in the crystal. It has been successfully shown that GaN can be grown with Ga terminated surfaces (known as Ga-polar GaN), which

PAGE 24

7 results in a smoother surface morphology and bett er quality films. The successful growth of Ga-polar GaN via MBE has a very narrow ma rgin of error when it comes to the exact growth parameters and to assist with the correct termination many growths of Ga-polar GaN start with a MOCVD GaN template, whic h has the right orientation for continued growth of the GaN film by MBE. Defects in GaN degrade device performa nce and longevity. Therefore it is important to have an understanding of the t ypes of defects present and the effects they produce in GaN. Native defects in GaN tend to be Ga or N vacancies, impurities, dislocations and/or interstitial atoms. On e non-destructive charact erization used to determine the types of defects and their e ffects is photoluminescence spectroscopy (PL). These defects cause different energy or color bands in the PL spectra. By monitoring the intensity changes and wavelength peak shifts in these bands, it is possible to identify the types of defects that are present. Knowing this information concerning the defects can allow for treatments pre or post GaN growth to reduce the defects. Common color bands that show up in GaN ar e the ultraviolet band (UVL ) at approximately 390-455nm (3.076eV3.257eV), blue ba nd (BL) at approximatel y at 438 to 442nm (2.88eV – 2.90eV), green band (GL) at 492 to 577nm( 2.5eV – 2.2eV), the much debated yellow band (YL) at 577 – 597nm (2.2eV – 2.3eV), and a red band (RL) at 688nm – 652nm (1.8eV – 1.9eV). Each of these bands corresponds to an energy transition within the band gap of gallium nitride and depending on the energy or band associated with the transition, pertains to a particular type of defect within the material. The ultraviolet band (UVL) has shown up in undoped GaN as well as Mg-doped GaN. In u-GaN the main candidates for

PAGE 25

8 shallow donors are SiGa and ON. The assignment of shallow acceptor has been identified using optical detected magne tic-resonance (ODMR) as SiN and in Mg-doped GaN, the shallow acceptors are MgGa and SiN. The UVL bands in each type of GaN are very similar to each other indicating the different shallow acceptors manifest themselves similarly in GaN as a transition from a shallow donor to a shallow acceptor. The blue band that occurs approxim ately 420nm to 490nm is observed in undoped, and Mg-doped GaN but only those grown via MOCVD or HVPE because the defect is not a native defect that is grown into the GaN film. The BL band has been defined in uGaN as a transition from a shallow donor to deep acceptors at low temperatures and at elevated temperatures from conduction band to deep acceptor. In p-GaN the transition was described as a transition from deep donor to the shallow MgGa acceptor. Quenching of the BL band occurs when holes from the acceptor level escape to the valence band and an observation of the BL band at low temperat ures such as 15K enables the determination of the vibrational characteristics of the defects in its ground state. The green band in GaN is usually coupled with the YL band and sometimes overshadowed by the intensity of the YL band. However, in high purity GaN a yellowgreen band can be observed at room temper ature. The shape and position of the band depends on the excitation intensity and exci tation energy and can always be deconvolved into two bands. It is presumed that the YL and GL band are related to two charge states of the same defect, presumably the VGaON complex. Many researchers have disputed the attribut ion of the YL band in GaN and are still undecided as to the exact orgins of the YL band. Although they have agreed that the YL band is not specific to a partic ular impurity defect, they have instead related the YL band

PAGE 26

9 to a native defect that is grown into the gallium nitride and is specific to a type of transition with the band gap. It is al so believed that in undoped GaN the VGaON complex is responsible for the YL band and that ON and CGa are shallow donors. It is still debated whether or not the YL band is a transition from the conduction band or shallow donor to a deep acceptor or as one researcher sugge sted [hoffman], a two stage process that involves a non-radiative captu re of an electron by a deep double donor followed by a radiative recombination between the electron at the deep donor and a hole at a shallow acceptor. Whichever transition is responsible, th us far no model fully describes the defect or defects responsible for th e YL band. What has been f ound about the YL band is that carbon doping and Si doping to a certain concen tration enhance the YL emission. It is believed that the carbon and silicon sit upon Ga vacancies. For Si, the initial enhancement comes from inadvertent increase in the concentration of the VGa–related acceptors such as VGaON and VGaSiGa facilitated by the shift of the Fermi level to higher energies during growth. The activation energy of the SiGa donor is 30meV. Although carbon does enhance the YL emission, the likelihood of a CGa formation is very low. The red band in u-GaN is the least studi ed and the least common of all the color bands that appear. The intensities are inde pendent of temperature, saturate at low excitation, and possess long lifetimes. The tran sition is known to be shallow donor or conduction band to a deep acceptor at low te mperatures and elevated temperatures respectively. In Ga-polar GaN two color bands dominate within the low temperature PL spectra, a red band (RL2) and a green band (GL2). Bo th tend to have very weak emission and quench above 100K. These bands are not to be confused with the other RL and GL bands

PAGE 27

10 mentioned. The bands are created by point de fects that are fairly uniform throughout the bulk GaN and tend to dominate high-resistivity GaN (no=1015 cm-3). Due to internal transitions in the defects, the Gaussian shap e of the bands indicate that in both cases the carrier is strongly localized at the defect providing a strong electron-phonon coupling. In summary, determining the color spectra of PL is a very useful way to thoroughly characterize all types of gallium nitride films and to determine what defects are grown in and what defects are induced due to doping, impurities, and irradiation. 2.4 Dielectrics 2.4.1 Ideal Dielectrics Ideal dielectric materials are perfect insula tors in which no mob ile charged particles are present. The ability to st ore an electric charge is calle d capacitance [3]. When an insulator is placed be tween the conducting plates of a capacitor, the capacitance can increase significantly. Insulators have posit ive and negative charges in the form of the atom nucleus and the electron cloud, but these charges are bound to th e atom or molecule and are not available for conduc tion [4]. Under the influence of an electric field, the nucleus and the electron cloud are di splaced to a form a dipole [1]. Figure 2-4 shows the dipole formation in the presence of an electric field. The total effect of an electric field on a dielectric material is known as polarization. The ability of the material to resist the polarization of charge is descri bed as the dielectric constant, which is the ratio of the permittivity of the material, i, to the permittivity of vacuum, v. = i/ v (2-1) The dielectric constant can also be relate d to the internal field created within the material and the external field applied, through Equation 2-2 [5].

PAGE 28

11 Einternal = E applied / (2-2) The polarization, P, of the ma terial is related to the di electric constant by equation 2-3, where is the strength of the electric field (V/m). It can be assumed from this relation that the polarization increases as the el ectric field strength increases, until all the dipoles are aligned such that P=( -1) o (2-3) There are various applications for dielectric materials. Passivation of high voltage junctions, isolation of devices and interconne cts and gate insulation of field effect transistors are a few that are relevant for this discussion. For a materi al to be a successful dielectric it must meet certain criteria. Desirable characteristics include chemical stability over the life of th e device, immobile charge tr aps (to avoid shorting and frequency limits) and a dielectric constant hi gher than that of the semiconductor (to avoid generating a high electric fiel d in the dielectric). The dielectric/semiconductor interface is also an important focus of research in the area of device proce ssing. The interface state density of carrier traps must be
PAGE 29

12 interface between the dielectric and substrate and current leakage in the oxide layer. Ideally the perfect dielectric would be a sing le crystal structure with the same symmetry and no lattice mismatch to GaN. The absence of lattice mismatch eliminates stress and defects at the interf ace. Realistically, using a dielectri c material with small lattice mismatch to the substrate should reduce high de fect densities at the interface. However, in some cases it may not be possible to e liminate a sufficient number of crystalline defects; thus the possibilities of using amorphous oxides will also be discussed. Using a single crystal layer of the prosp ect oxides should in principle provide a high quality layer. However, ev en the small lattice mismatch of the two prospect oxides can still cause defects in the interface. Thes e defects can propagate through the entire oxide creating traps or leakage paths. One pr oblem is finding the right growth conditions that will produce a single-crystal and not a poly-crystalline structure. Polycrystalline layers at the interface could result in gr ain boundaries allowing high leakage current through the dielectric. In an amorphous arrangement the at oms are randomly distributed on the substrate. This eliminates the problem of strain and dislocations at the interface and in the dielectric. A truly amorphous laye r would also stop any shortening of current through the oxide, because the electrons or hol es would have no defect paths to follow. One potential problem is the possibility for form ation of crystallites in the layer. At high operating temperatures, the crystallites may gr ow and coalesce to form grain boundaries and a polycrystalline-like structure, result ing in a very leaky and thermally unstable dielectric. Each type of oxide laye r has advantages and disadvantages. 2.4.3 Present State of Dielec trics for Gallium Nitride Several dielectrics have been tested for use on wide bandgap GaN semiconductor devices (Table 2-1). Gallium nitride resear ch has resulted in long-lifetime, room

PAGE 30

13 temperature operation of photonic devices. Thes e include LEDs that cover the visible spectrum, laser diodes in the blue and blue-green regime and UV detectors. These devices are just recently reach ing production with problems still to be solved in the fields of n-Ohmic and p-Ohmic contac ts, p-type doping issues, Scho ttky contacts, and dielectric materials. Also, with the lack of availabil ity of high-quality GaN s ubstrates, research an epitaxial growth and substrate selection is still ongoing. From mate rial and processing advances learned from the phot onics research, high-power an d high-temperature switches have been realized. The following summarizes some of the dielectric s research to date. 2.4.3.1 Silicon oxide on GaN Silicon oxide is a very attractive choice for a dielectric material since it is has been well studied and the processing is well esta blished. Silicon oxide deposited by plasma enhanced chemical vapor deposition (PECVD) [6-9] has been reporte d to give interface state densities on the order of low 1011 eV-1cm-2. Silicon oxide deposited by electron beam (EB) evaporation on GaN has s hown an interface state density of 5.3x1011 eV-1cm-2 [8]. After annealing the EB deposited SiOx at 650C the valance band offset was measured to be 2eV and the conduction band offs et was measured to be 3.6 eV [10]. The EB evaporated SiOx shows a silicon rich stoichiometry when compared to the PECVD SiOx. There are several inhere nt problems with SiOx as a dielectric material for wide bandgap semiconductors. The high Dit of SiOx is due to uncontrolled oxidation of the surface [11]. The most significant limitation is that SiOx has a dielectric constant ( ) of 3.9, which is significantly lower than that of GaN ( = 8.9). This will create a very large

PAGE 31

14 electric field in the dielectric, leadin g to further breakdown and increased power consumption. 2.4.3.2 Silicon nitride on GaN The Dit obtained from PECVD silicon nitride [8,9] has been reported as 6.5x1011 eV-1cm-2. The Si3N4 MISFET ,however shows poor performance. Electrical measurements showed the MISFET structure ha d a large flat band voltage shift (3.07 V) and a low breakdown voltage (1.5 MV/cm). There was no microstructural analysis performed on the deposited Si3N4 films. SiN deposited by ECR plasma CVD showed a Dit of 1x1011cm-2eV-1 but had excess leakage current due to the small conduction band offset [11]. The ECR-CVD MIS diode showed a Dit of 4x1011 cm-2eV-1, fixed oxide charge of 1.1x1011 cm-2 and breakdown 5.7 MV/cm with a di electric constant of 6 [12]. A unique dielectric structure of SiO2/Si3N4/SiO2 (ONO) was reported to have breakdown field strength of 12.5 MV/c m for temperatures as high as 300 C.20 The ONO structure was deposited by jet vapor deposition to a thicknes s of 10 nm /20 nm /10 nm. The multiplayer structure does allow for unique engineering of a dielectric, but multiple interfaces can lead to an extremely large number of inte rface state traps and increased processing. The Dit for the ONO structure was re ported to be less than 5x1010 eV-1cm-2 with breakdown fields great er than 12 MV/cm [13]. 2.4.3.3 Aluminum nitride on GaN Aluminum nitride deposited by MBE a nd MOCVD has been used to create MISFET devices and insulated gate heterostru cture field effect transistor (IG-HFET) devices [14,15]. The AlN MI SFET structure grown at 400 C was polycrystalline. From x-ray reflectivity measurements, the AlN/GaN interface showed a roughness of 2.0 nm.

PAGE 32

15 This may be due to the polycrystalline nature of the film or from intermixing of the AlN and GaN at the interface. The dielectric breakdown fi eld was calculated to be 1.2 MV/cm. The AlN IG-HFET structure was grown at 990 C, forming a single crystal film of 4.0 nm. This device ope rated in depletion mode and had a pinch-off voltage of 0 V. Hexagonal aluminum nitride has a 2.4% la ttice mismatch with hexagonal GaN on the (0001) plane. The 4.0 nm film thickness is great er than the critical thickness allowed for elastic deformation leading to threading disloc ations forming from plastic deformation. Single crystal AlN and polycrystalline Al N films suffer from defects and grain boundaries that cause shorting. 2.4.3.4 Gallium oxide on GaN Like Si, gallium nitride forms a native oxide This oxide has been considered as a dielectric material, like the native oxide of silicon. Thermal oxidation has been studied using dry [16,17] and wet [18] atmosphere s. Dry oxidation of GaN epilayers at temperatures below 900 C showed minimal oxidati on. Dry oxidation at 880oC for 5 hours produced 1110 nm of -Ga2O3 with a Dit of 1x1010 eV-1cm-2 and showed inversion [19]. At temperatures above 900oC, a polycrystalline monoclinic Ga2O3 formed at a rate of 5.0 nm/hr. This oxidation rate is too slow to be viable as a processing step. Wet oxidation of GaN also forms polycrystalline monoclinic Ga2O3, but at a rate of 50.0 nm/hr at 900 C. From cross-sectional transmi ssion electron microscopy, the interface between the oxide and the GaN was found to be non-uniform. Scanning electron microscopy showed that both films were rough and faceted. Electrical characterization of the oxide showed a dry oxide dielectric field strength of 0.2 MV/cm and a wet oxide dielectric field strength of 0.05 to 0.1 MV/cm. The microstructure formed from this

PAGE 33

16 process was shown by XRD to be a high temperature hexagonal phase [20]. Ga2O3, formed by this method passivates the surface [20-23] and has a Dit of 1011 eV-1cm-2 for GaN MOS. A negative oxide charge as well as high capacitance and reduced reverse leakage where shown for thicker oxides gr own by PEC [21]. Using PEC and a HeCd laser, a low reverse leakage current of 200 pA at 20Vm was achieved. For this oxide the forward breakdown, Efb, was 2.8 MV/cm and the reverse breakdown, Erb, was 5.7 MV/cm with a Dit of 2.53 x 1011 cm-2eV-1. The dielectric constant of Ga2O3 grown under these conditions was estimated to be10.6 [24]. With amorphous GaO deposited by PEC, low leakage currents of <10 x 10-8 /cm at –15 V have been measured. Using Ga2O3 as both gate dielectric and passiva tion layer, a breakdown field of 0.4 MV/cm was observed. The bandgap, Eg, of Ga2O3 was measured to be 4.4 eV [25]. 2.4.3.5 Silicon dioxide on gallium oxide on GaN Depending on the growth technique, the in terface between the SiO and the GaN can vary drastically. When SiO is deposited by RPECVD, a parasitic subcutaneous layer of native gallium oxide is grown on the GaN surface. This layer has been shown to have a direct effect on the device pe rformance. When the thickness of the initial GaO layer is controlled by a pre-oxidation step the device characteristics improve markedly. Remote Plasma Assisted Oxidation first followed by RPECVD gave a lower Dit and a smaller flat band shift over the RPECVD of the SiOx alone [26]. Real and ideal CV curves are nearly identical [27]. After an anneal in an RTA for 1 min at 900oC in Ar the Dit is 2-3x1011 cm-2 [28,29]. Another group used a sim ilar oxide growth technique and measured a Dit of 3.9x1010 eV-1cm-2 and a low leakage current [30, 31].

PAGE 34

17 2.4.3.6 Gallium gadolinium oxide on GaN Due to the recent success of gallium gadolinium oxide (GGG) as a dielectric in GaAs MOSFETs [32-37], attention has turned toward this as a dielectric material for GaN. The GGG dielectric was deposited on a GaN epilayer by EB evaporation of a single crystal GGG source [14]. Th e substrate temperature was 550 C. The interface roughness was calculated to be 0.3 nm fr om x-ray reflectivity. Metal oxide semiconductor (MOS) capacitors were formed and tested. A breakdown field of greater than 12 MV/cm was estimated. More recently, the thermal stability of the film and the interface has been proven to temperatures as high as 950 C and operation of a depletion mode MOSFET has been performed at temperatures up to 400 C [38]. The EB evaporated GGG stoichiometry is heavily de pendant upon the substrate temperature. Changes in temperature lead to changes in th e stoichiometry [39].This limits the available microstructure obtainable within th e stoichiometric limits of GGG. 2.4.3.7 Gadolinium oxide on GaN GaN based MOSFETs have been made that used a stacked gate oxide consisting of single crystal gadolinium oxide and amorphous SiO2 [40]. The gadolinium oxide provides a good oxide /semiconductor interface and the SiO2 reduces the gate leakage current and enhances oxide breakdown vo ltages. The dislocations in the Gd2O3 film limit the breakdown field that can be sustained in the dielectric. 2.4.3.8 Scandium oxide on GaN Scandium oxide grown by MBE has been us ed as a gate dielectric and passivation layer for GaN based devices. Scandium oxi de has the bixbyite crystal structure, a reasonable band gap of 6.3 eV and a lattice mismatch to GaN of 9.2%. The scandium

PAGE 35

18 oxide was grown by MBE using an RF oxyge n plasma, substrate temperature of 650 C and an effusion cell temperature of 1130 C. This oxide showed inversion when used in a GaN gated diode [41]. This oxide was also grown under the same conditions except the substrate temperature was 100 C, which resulted in an interface state density of 5x1011 eV-1cm-2 [42]. Scandium oxide has also been used as a fi eld passivation layer fo r GaN devices. It has been shown to reduce the reverse leakage current and increase the fT and fMAX. The passivation films have been grown w ith a substrate temperature of 100 C and a cell temperature of 1130 C [43]. Scandium oxide has bett er long-term stability than SiNx as a passivation film for GaN based HEMTs. It has been shown to dr amatically reduce the gate lag problems due to surface states on AlGaN/GaN HEMTs [44]. 2.5 Epitaxal Growth The term “epitaxy ” is Greek in origin, coming from the words “epi” meaning upon, and “taxis” meaning to arrange. Epitaxy is the process of controlled growth of a high-quality crystalline or amorphous layer of material on a substrate. There are two categories of growth for epitaxy: homoep itaxy and heteroepitaxy. Homoepitaxy is the growth of a crystalline layer that is the same as the substrate material while heteroepitaxy is the growth of a layer that differs from th at of the substrate [1]. The growth method used in this study is Molecula r Beam Epitaxy (MBE). It is important to select growth parameters that will produce the highest quality epitaxial layer for the GaN semiconductor, in order to reduce crystalline defects that would de grade the performance of the semiconductor device.

PAGE 36

19 2.5.1 Molecular Beam Epitaxy The growth of the epitaxy layer takes pl ace due to reactions between the molecular beams coming from the sources and a crystal line surface held at suitable temperatures. The molecular beams are produced from solid el ement at sources heated in effusion cells. Growth begins when different atomic speci es are absorbed on the substrate surface and migrate to form the deposited layer. A comm on in-situ growth-monitoring technique is electron diffraction, commonly known as RH EED (Reflection High-Energy Electron Diffraction). In this method, high-energy elec trons are diffracted off the growing surface and produce an image on a screen on the opposit e side of the chamber [45] (Figure 2-5). Although MBE has the capability for growing complex multilayers where precise control of dopant concentration, layer thickne ss and interface abruptness are required, there are also serious concerns. For example, due to the UHV environment sensitivity to contamination and replenishment of sources, lo ng periods of system downtime can ensue. 2.5.2 Substrate Preparation Before any epitaxial growth, the substrates received an ex-situ and an in-situ surface treatment to remove any contaminati on and the native oxide. The surface of the semiconductor must be as clean and planar as possible to ensure high quality dielectric film deposition. Surface contaminat ion leads to impurities at the dielectric/semiconductor interface, which ultimat ely result in the form ation of interface traps. Since gallium nitride substrates are not currently available, gallium nitride grown on sapphire wafers oriented <0001> were used. These will be referred to throughout this work as GaN substrates. Two different type s of growth of the GaN substrates were employed in this work, MBE and metal-orga nic chemical vapor deposition (MOCVD). The MBE GaN substrates were grown in hous e (referred to as UF MBE GaN) and the

PAGE 37

20 MOCVD substrates were either grown in house (referred to as UF GaN) or by Uniroyal Optoelectronics (UOE). From AFM, ther e is a large differen ce in surface roughness between the two types of GaN. The MOCV D substrates are 13 nm RMS roughness and the MBE substrates are ~6 nm RMS roughness (Figure 2-6). The GaN substrates received an ex-situ treatment consisting of a 3-minute etch in (1:1) hydrochloric acid: water, followed by a DI rinse and blown dry by nitrogen. This was use to remove any organic residue from the surface. Then a 25-minute exposure to ozone produced by an ultraviolet lamp in a UVOCS UVO Cleaner model number 42-220 was used to oxidize the carbon on the surface and create a thin native oxide layer. Next, the substrates received another etch in buffe red oxide etch for 5 minutes to remove the native oxide. This is shown by observing the RHEED pattern produced from the surface. The RHEED pattern produced by the native oxide is more diffuse than the pattern produced by the buffered oxide etched surface (Figure 2-7). The GaN substrates were then mounted to molybdenum blocks using indium solder, and then loaded into the load-loc k of the MBE. Room temperature RHEED measurements showed a reasonably clean (1x1) surface (Figure 2-7). Two different crystal directions are observed in RHEED to create a more complete understanding of the surface. Here, the <1-100> and the < 11-20> directions are observed. An in-situ thermal treatment was employed to further remove any oxide or contamination left on the surface. The substrates were heated to 700oC in vacuum and no overpressure of nitrogen was used. The RHEED patterns reco rded at this temperature in dicate a sharp (1x3) pattern (Figure 2-8).

PAGE 38

21 2.6 Types and Effects of Radi ation on Microelectronics In the past decades most radiation testing has been performed on Si devices. Currently there is very limited informa tion on radiation effects on GaN devices. The objective is to investigate the stability of the oxide, the interface between the oxide and semiconductor, and the operational ch aracteristics of the MOS capacitor. In order to understand the relevant applications for radiation hard devices one must first understand the origins and effects of the space radiation and terrestrial radiation environment. There are various types of radiation that affect electronic devices such as gamma rays, x-rays, protons, neutrons and othe r subatomic radioactive particles. Much of this radiation comes from solar cosmic rays, galactic cosmic rays that originate outside the solar system, and other planetary envir onments. The earth’s magnetic field shields a region of near earth space from these partic les but they easily reach polar regions and high altitudes such as geosynchronous orbit ( 35,800km) in which satellites travel [46,47]. Around the earth lies a radiation environment known as the Van Allen belts. These belts are compromised of trapped subatomic partic les that can affect electronic devices operating at high-altitudes. Gamma rays are electromagnetic radiat ion emitted by radioactive decay. These rays are shorter in wavelengt h than ultra violet and are far higher in energy. For experimental use, common sources of gamma rays are Cobalt 60 and Cesium 137. Effects of gamma rays tend to occur over long te rms and can cause changes in electrical operational characteristics. X-rays are created from high-speed el ectrons that after impinging on a metal release energy in the form of an electro magnetic wave. Unlike gamma rays, x-rays consist of a mixture of different wavelengths. In addition to origina ting from solar flares

PAGE 39

22 and other cosmic occurrences, x-rays are al so produced by nuclear explosions. After a nuclear explosion, much of the kinetic ener gy of the fission fragments is converted to internal excitation and radiation. Some 70-80% of the total energy of this material is emitted as thermal electromagnetic radiation in the soft x-ray region. This can produce ionizing effects on MOS devices. Radiation of this type can be f ound at satellite orbit altitudes as well. Neutrons are also a concern with radiat ion environments. Neutrons are released following nuclear fission and can travel la rge distances even at sea-level and at atmospheric pressures. The neutrons can produ ce recoil protons by el astic collisions in hydrogen rich materials, therefore ionizing th e material or electronics of interest. Each of these types of ra diation can damage or upset electronic devices but in comparison to the amount of protons in the environment they make up a small percentage. Generally, protons tr apped in the earth’s magnetosphere have energies up to 800 MeV. They primarily originate from galactic cosmic rays, which come from outside the solar system. These protons are concen trated in a small area known as the South Atlantic Anomaly. The radiation belt dips into the earth’s a tmosphere due to the earth’s tilt on its axis. This causes concern for electronic devices that will operate in or near the region. Hence the focus of this study, eff ects of proton radiation on gallium nitride devices. Proton radiation causes ionization in electronic device s. This involves creating electron-hole pairs, Frenkel defects, and bulk semiconductor defects [48]. Proton radiation can also cause traps to form at th e interface of the oxide and substrate. These various radiation-induced defects can cau se current leakage through the oxide, displacement or removal of semiconductor atoms, build up of trapped charge and shifts in

PAGE 40

23 operational parameters of the electronic device. Previous st udies on silicon devices have shown radiation resistance under the men tioned environments. Gallium nitride is expected to be even more radia tion resistant or rad-hard than silicon. In the course of this study 40MeV(+2%) and 10MeV (+10%) energi es of protons will be examined at 5x109cm2 total particle dose. Although lower en ergy protons can be blocked by shielding, it is useful to use lower energy protons in orde r to look at the type of damage caused by radiation to the device or material. As tec hnology improves, device features get smaller and use lower energy to operate. This can cause sensitivity to various impinging particles, which can damage the functionality of the device. The key to reducing the radiation sensitivity of the device may rely on finding an oxide layer that can remain virtually unaffected while protecting the underlying semiconductor material and interface. In addition to permanent degradation of th e device, an electrical device can also experience a number of single event effects (SEE) due to total dos e ionization of the material or device. Soft errors are single even t upsets (SEU). Events of this type can be corrected by reprogramming or resetting the device. This is usually apparent in memory circuits. It can also result in performance de gradation if the error rate is too high. Hard errors, which are not correctible, include si ngle event latchup (SEL) burnout (SEB), and gate rupture (SEGR). If a hard error occurs a circuit elem ent can be physically damaged. 2.7 Previous Radiation Studies In recent years, several studies have been done on radiation of various GaN devices. Radiation types have included gamm a rays, electrons and low-energy protons. Gamma irradiation at total fluences of 600 Mrads was studied by Luo et al. [49]. At this high dosage the AlGaN/GaN HE MT showed a 45% change in the

PAGE 41

24 transconductance, gm, for varied gate lengths and wi dths. At a lower dose of 300 Mrad, minimal changes in the electrical characteristics were observed. Look et al. [50] conducted an electron radiation study. Energies ranging from 0.7 to 1 MeV were used to induce removal of n itrogen or gallium atoms. Fluences ranging from 1 to 7 x1016 cm-2 were noted. They concluded that nitrogen Frenkel pair formation was occurring, resulting in the formati on of shallow donors and shallow or deep acceptors at the same rate. Look’s group also concluded that annealing produced Frenkel pair recombination. This work also supports th e donor nature of N vacancies in the GaN. The effect of proton radiation on AlGaN/InGaN/GaN LEDs was reported by Osinski et al. [51]. This group used a beam energy of 2 MeV with a total fluence of 1.68 x 1012 protons/cm2. They found a 40% reduction in the output power of the LED even though the I-V curves of the devices showed ve ry little change. The optical properties of the irradiated area almost returned to norma l when the sample temperature was lowered to ~15 K. Their conclusion was that the proton dosage did not degrade the single quantum well LEDs and the optical prope rties remained practically unchanged. Another group, Khanna et al. [52], studi ed photoluminescence of proton irradiated GaN to determine its radiation resistance. Proton beams with an energy of 2 MeV and varying fluences were used. As expected, they found that GaN was much more radiation resistant than GaAs. This was determined from PL intensity levels. A reduction in the intensity of the dominant peak in the PL sp ectrum of irradiated GaN was noted, implying that midgap states were created. They attribut ed the intensity loss as a result of trapped carriers at radiation-induced defect sites.

PAGE 42

25 In a study of annealing behavior of pr oton radiated AlGaN/GaN HEMTs, they found remarkable recovery of the HE MT DC performance. A fluence of 1x1014 cm-2 at an energy of 1.8 MeV was used. After radiati on of the HEMT the saturation of the I-V curves was reduced from 260 mA/mm to about 100 mA/mm. As with other studies the reduction was attributed to radiation induced traps resulting in removal of free carriers. Transconductance of the device was also reduced from 80 mS/mm to 26 mS/mm. However the study showed a gradual increase of electrical characteri stics with increasing annealing temperatures. At a temperature of 800oC, the saturation and transconductance returned to 220 mA/mm and 56 mS/mm respectively [53]. An experiment conducted by Emtsev et al., studied radiation i nduced defects of ntype GaN and InN. 1.0 to 1.5 m thick layers of hexagonal n-GaN and InN were grown by MOCVD and plasma–assisted MBE techniqu es. After pre-radiation characterization by XRD and Raman spectroscopy, the samples we re radiated with protons of 150keV and then annealed at 50oC or 100oC for 20 minutes in nitrogen. For the InN, the group found that the increase in electron concentration was most likely due to the production of radiation–induced def ects with the shallow donor states. Since the production rate did not change over a wide range of proton dosage, 1e1015 to 1e1016cm-2, the defects are believed to be native defects that can be attributed to immobile nitrogen vacancies in the InN. Annealing temperatures up to 100oC show no change in electron concentration or electron mobility but did show a pronounced de crease of 30% in electron concentration between temperatures of 250-300oC. At elevated temperatures the annealing behavior becomes complicated in the sense that a reve rse annealing stage of electron concentration and mobility takes place at approximately 400oC. At 500oC little change in either

PAGE 43

26 concentration or mobility was noted but the parameters were higher than the nonirradiated layers of InN. N-GaN: Si showed a substantial decrease in concentration of charge carriers. It was found that the el ectron removal rate was dependent on doping level of the GaN. Af ter annealing to 200oC, there is a noticeable decrease in electron mobility but no significant changes in electr on concentration. At temperatures of 300400oC the mobility continues to drop but at 600oC the electron concentration and mobility recover substantially. It is noticed that the mobility of the charge carriers becomes even larger than the initial measured mob ility of the n-GaN. Based on this study the group concluded that in InN, nitrogen vacan cies were the likely cause of the increase of free electrons in the irradi ated InN. For n-GaN they conc luded that the production rate of native defects appears to be Fermi-level de pendant and that two ma in recovery stages of electron concentration were found at intervals of 300-400oC and 500-600oC [54]. Gaubas et al., studied the radiation effects of semi-insulating layers of GaN grown on bulk n-GaN/sapphire substrates. The sample s were irradiated w ith 10keV X-ray dose of 600Mrads and 100keV neutrons with fluences of 5e1014 and 1016 cm-2 respectively. A set of wafers and diode structures were irra diated then characterized post radiation using Photoluminescence spectroscopy (PL), non-inva sive microwave absorption (MVA) and contact photoconductivity (CPC). The MVA method is based on a pump-probe technique with optical excitation and mi crowave absorption by the free carriers, the CPC relies on measuring the photocurrent decays. It was found that the radiation induces an increase in the non-radiative trap density, which resulted in a significant decrease in the PL intensity of the blue, yellow and UV bands. The effects of the disorder cause d by the radiation

PAGE 44

27 manifested themselves in the long tail CPC measurements as well as MWA decays with a time stretching factor of .07 [55].

PAGE 45

28 Table 2-1 Properties of Dielectrics Pr eviously Studied for Use with GaN Material Bandgap (eV) Dielectric ( ) Melting Point (K) AlN Al2O3 Ga2O3 Gd2O3 Ga2O3(Gd2O3) SiOx Si3N4 MgO Sc2O3 6.2 5.75 4.4 8.5 4.7 8-9 5.0 7.3 6.3 8.5 12 10 11.4 14.2 3.9 7.5 9.8 11.4 3273 2319 2013 2668 2023 1993 2173 3073 2678

PAGE 46

29 Figure 2-1 Typical MOSFET n-type Substrate Insulator Source Drain Gate

PAGE 47

30 A B Figure 2-2 Cross-section illustration of a enhancement mode MOSFET. A) At VG=0 the device is “OFF” B) At VG<0 the device is “ON”. Notice the conduction channel of electrons created with th e application of positive gate voltage. n-type regions n-type regions p-type Substrate Source Drain Oxide Gate VG = 0 p-type Substrate Source Drain Oxide Gate VG < 0

PAGE 48

31 Figure 2-3 Cross secti on of a MOS capacitor. Figure 2-4 Dipole formation in the presence of an electric field. Semiconductor Substrate Dielectric Metal Metal + + E +

PAGE 49

32 Figure 2-5 Sketch of Riber MBE used for oxide growth (after Gila) ECR PLASMA SOURCE SOLID SOURCES RHEED Gu n Load/lock Bufferchamber

PAGE 50

33 A B Figure 2-6 AFM images of as received Ga N. A) MOCVD GaN from Epitronics B) as received MBE GaN from SVT.

PAGE 51

34 A B Figure 2-7 Examples of GaN surfaces befo re growth. A) The UV-ozone treated surface of GaN. B) Buffered oxide etched surface of GaN. A B Figure 2-8 Photos of RHEED indicating a (1x3) pattern. A) <11-20> crystal direction. B) <1-100> crystal direction.

PAGE 52

35 CHAPTER 3 EXPERIMENTAL APPROACH 3.1 Alternative Dielectrics Because of the various limitations of the pr eviously studied dielectrics, alternative dielectrics are being actively sought. Scandium oxide, magnesium oxide and the ternary combination of magnesium calcium oxide s how promise as candidate dielectrics. MgO alone as a gate insulator has proven to have a low Dit, but is thermally and structurally unstable in normal atmosphere. The possible so lution to this problem would be to cap the MgO with a more stable oxide layer or comb ine the magnesium with another dielectric material add stability to the oxide. Hence, the purpose of using a magnesium ternary oxide capped with thermally a nd structurally stable scandi um oxide as dielectric. The reasons for selecting these pa rticular oxides are as follows: Lower lattice mismatch to the GaN semiconductor Wide bandgap Thermal stability High dielectric constant Desirable band alignment with GaN Each of the proposed dielectrics have bandgaps that are substantia lly larger than the bandgap of GaN. In addition to withstanding high operating temperatures, the dielectrics should also be able to withstand irradiat ion of protons that are found in caustic environments of space and low earth orbits.

PAGE 53

36 3.2 Oxide Growth Parameters 3.2.1 Scandium Oxide Growth The Sc2O3 samples were grown using the Riber 2300 MBE equipped with a RHEED system. Scandium oxide exists in a bixbyite structure and was grown using a standard effusion cell filled with Sc opera ting at temperatures from 1130C to 1170C [56,57]. Oxygen was supplied from a RF plas ma source which, was kept at 200 watts forward power. The chamber pressures ranged from 1x10-4 Torr to 5x10-4 Torr. The substrate temperature was varied from 100 C to 600C. Higher growth temperatures produce larger grain sizes. Electron micros copy shows a smooth interface between the Sc2O3 and the GaN substrate. 3.2.2 Magnesium Oxide Growth The magnesium oxide samples were grown from 99.99% pure magnesium and Knudsen cell temperatures ranging from 350 C to 400 C. Substrate temperatures between 100 C and 340 C were used. The oxygen was provided by an Oxford RF plasma source at 300 W forward power. Oxygen pressure was varied from 8x10-6 up to 1x10-5 Torr. In all cases, the sample rotation was kept at 15 rpm. 3.2.3 Magnesium Calcium Oxide Growth The magnesium calcium oxide samples were grown from the same Mg sources as that used for the MgO with the addition of 99.999% pure calcium, Ca, from another Knudsen cell with temperatures ranging from 450 C to 500 C. Substrate growth temperatures were between 100 C and 300 C. Two growth methods were used: 1) continuous where all shutters open at once a nd exposed to the subs trate and 2) digital alloying where alternating layers of MgO a nd CaO are deposited. Changing the flux of

PAGE 54

37 the metal sources during a continuous growth or changing the timing of the shutter sequences during digital growth varied the co mposition of the film. Oxygen pressure was held at 8x10-6 Torr and the RF plasma source. As in the MgO growth, the sample rotation was kept at 15 rpm. 3.3 Materials Characterization 3.3.1 Auger Electron Spectroscopy Auger Electron Spectroscopy (AES) uses a focused electron beam to create secondary electrons near the surface of a solid sample [58]. Auger electrons are able to characterize the elemental composition and at times, the chemistry of the surface of samples. When combined with ion sputteri ng to gradually remove the surface, AES can similarly characterize the sample in depth, allowing microanalysis of three-dimensional regions of the solid samples. Auger is normally non-destructive except during depth profiling and for materials which are sensitive to the e-beam. Th e main use of AES is to discover the elemental composition of inor ganic materials or interface compositions. (Figure 3-1) 3.3.2 Atomic Force Microscopy Atomic Force Microscopy (AFM) is a real space imaging technique that can produce topographic images of a surface with atomic resolution in all three dimensions [58]. Atomic Force Microscopy is a very powerful imaging system since it can study insulators, semiconductors, conductors, and tr ansparent as well as opaque materials. Surfaces can also be studied in liquid or in ultra high vacuum and the system uses a sharp tip mounted on a flexible cantilever. When th e tip comes within a few angstroms of the sample’s surface, the repulsive van der Waals forces between the atoms on the tip and

PAGE 55

38 those on the sample cause the cantilever to deflect. Only w ith an unusually sharp tip and a flat sample is the lateral reso lution truly atomic. (Figure 3-2) 3.3.3 Transmission Electron Microscopy One of the most powerful microstructu ral analysis techni ques available is transmission electron microscopy (TEM). Fr om TEM, not only can the microstructure of an epitaxial film be determined, but also deta iled analysis of defect s in the film, atomic imaging of the interface, and an accurate cal culation of lattice spacing is determined. TEM uses a beam of electrons that pass through and interact with a very thin sample to form an image on the other side of the samp le (Figure 3-3) [58]. The interactions between the atoms in the sample and the electr ons produce the contrast seen in the image. One of the major drawbacks of TEM is the sample preparation required to obtain the images. The sample must be cut, polishe d, and thinned to electron transparency (~100 nm) via hand polishing and ion beam milling or by using a focused ion beam (FIB) system. This is especially difficult for the nitride materials due to their hardness. The FIB used to make these sample was a FE I Strata DB (Dual Beam) 235 FIB. A JOEL 200CX operating at 200 keV was used for f ilm analysis and JO EL 2010FX operating at 400 keV was used for high-resolution analysis of the interface. 3.3.4 Scanning Electron Microscopy Scanning Electron Microscopy (SEM) is anot her technique used to give detailed material characterization. It uses a rastered electron micros cope to highly magnify the image of the surface of the ma terial [58]. The SEM works by scanning a fine probe of electrons over the surface of the specimen under vacuum conditions. As electrons penetrate and react with the surface, an emission of electrons or photons from the surface is given off. An appropriate detector can then collect the emission and the output can be

PAGE 56

39 used to modulate the brightness of a cathode ray tube (CRT), whose inputs are in sync with the voltages rastering the electron beam. The image is produced on the CRT and every point the beam strikes on the sample is mapped directly onto a corresponding point on the screen. (Figure 3-4) 3.3.5 X-Ray Reflectivity X-ray Reflectivity (XRR) is a non-destructiv e technique used to study the structure and the organization of materials, which ar e grown as thin films. The range of XRR includes the submicronic and atomic scales [59] It can be used to determine the electron density profile (EDP) and the roughness of interfaces. By measuring the diffuse x-ray scattering, it can show the in terfacial roughness between succ essive layers. The typical wavelength of the x-rays are 0.1 nm, which is a very high frequency. As a consequence, the x-ray interaction with matter can be described by an index of refraction, which characterizes the change of direction of the x-ray beam when passing from air to a material. The reflected intensities will be confined in a direction symmetrical from the incident x-rays beams. This technique is also very sensitive to a ny defects of flatness, which makes rough surfaces undesirable when performing this technique. The main advantages of XRR are the abilities to de termine the surface and interface roughness, the layer thickness, EDP and the structur al arrangements of complex films. 3.3.6 Photoluminescence Photoluminescence (PL) measures physical and chemical properties of materials by using photons to induce excited el ectronic states in the material and analyzing the optical emission as the states relax [58]. The spect ral distribution and time dependence of the emission are related to electronic transition probabilities within the sample, and can be used to provide qualitative and sometime quantitative information about chemical

PAGE 57

40 composition, structure (i.e. interfaces, bonding, di sorder), impurities, and energy transfer. This technique will allow useful observation of radiation effects to the interface between the dielectric and semiconductor. Figure 3-5 show s a setup of the PL system that is used in this study. A helium-cadmium laser is us ed as the excitation source and scanning parameters were from 340nm to 800nm. 3.3.7 Hall Effect The Hall Effect is used to determine car rier concentration of the GaN substrate before and after thermal anneals and proton i rradiation. The hall effect is a result of a magnetic field being applied perpendicular to the direction of the motion of charged particles. The magnetic field exerts a force on the current and causes the charged carriers in the current to split and ga ther at polar opposite sides of the sample. This enables the calculation of majority and minority charge ca rriers. Any change in the concentration in the carrier numbers can also be observed. The Hall voltage VH is represented by VH = (RH Ix Bz)/ d (3-1) Where RH is the Hall coefficient, Ix is the current in the x-direction, Bz is the magnetic field in the positive z-direction and d is the thickness of the sample [4]. 3.3.8 Current Voltage Analysis A Hewlett Packard semiconductor parameter analyzer, HP4145B, was used to take I-V measurements of the MOS devices. The uppe r and lower current limits were set and swept from negative to positive. The range of the voltage was increased incrementally until forward and reverse breakdown were obs erved. Figure 3-6 depicts an ideal I-V curve for MOS devices. Note the sharp upward slope of the curve. The reasoning for this is “turn on” of the device. Once enough volta ge is applied to th e gate, current is conducted across the dielectric. The breakdown field strength is given by Equation 3-2.

PAGE 58

41 EBD = V/t (3-2) Where t is the thickness of the oxide, V is the voltage at the compliance current, normally 1 mA/cm2. A low breakdown field of the oxide is undesirable in this study, since it indicates weakness of the dielectric material. 3.1.9 Capacitance – Voltage Analysis Capacitance-Voltage measurement is a critic al analysis of elec trical devices. C-V gives information about the fixed oxide charge interface state density, border trap density and mobile ion density through graph plots or mathematical computations. All C-V data was obtained using an automated HP4284A LCR connected to a LabVIEWTM base PC. The LCR meter supplies a volta ge signal of superimposed analog current (AC) and direct current (DC). Ideal C-V curves for n-type MOS capacitors are shown in figure 3-7. The C-V curves are frequencyindependent in a ccumulation and depletion but at the on set of inversion the curves become strongly fre quency-dependent. Mini mizing the interface state density, Dit, is also an important aspect of diel ectric materials. A high interface state density of an oxide negates its usefulness as an insulating material for the semiconductor. 3.2 Diode Fabrication Once the oxides were epitaxially deposited by MB E, the structures were fabricated into MOS diodes. Scandium oxide was etched usi ng either an Inductively Coupled Plasma (ICP) or using a hot wet etch of H2SO4 for 6mins. Magnesium Oxide was etched using a 2% solution of phosphoric acid and DI H2O for 20 seconds. The Magnesium Calcuim Oxide was etched using the same 2% Phosphoric Acid/DI H2O solution. Ohmic contacts consisted of 200 Ti/700 Al/400 Pt/1000 Au and were deposited by e-beam evaporation. Gate contacts were Pt/Au, 200 and 1000 respectively. No post anneal

PAGE 59

42 was performed after deposition of the contac ts. The finished MOS capacitors are shown in figure 3-8. 3.3 Proton Radiation Setup and Facility For proton radiation the samples were taken to the Texas A&M Cyclotron. The oxide and device samples were radiated at doses of 5x109 cm–2 at energies of 10MeV and 40MeV under a vacuum in the 10-4 Torr range. The chamber takes at least six minutes to pump down before radiation can start. The samples we re affixed to glass slides and then placed on the target mounting in the testing chamber as shown in figure 3-9. The target mount frame can be adjusted in x, y, and z direc tions and rotated by computer control. The size of the exposed area is controlled by a shut ter like aperture that can be adjusted horizontally and vertically to insure beam isolation to the desired target area. Beam uniformity and dosimetry are monitored by an array of five plastic scintillators located upstream from the targ et chamber. Beam uniformities of 95% or better can be achieved. Outside of the testing chamber, an electronic diagnostic setup is available for data acquisition, in-situ or ex -situ testing (dependi ng on the device) and quick target changes (Figure 3-10).

PAGE 60

43 Table 3-1 Proposed Oxide Properties. Bandgap (eV) Dielectic Constant ( ) Melting Point (K) Mistmatch to GaN Sc2O3 6.3 11.4 2678 9.2% MgO 7.3 9.8 3073 -6.5% MgCaOa 7.4 10.8 2370 GaN 3.4 9.5 2773 a. composition of approx 50/50

PAGE 61

44 Figure 3-1 Auger Electr on Spectroscopy set-up.

PAGE 62

45 A piezoelectric scanner which moves the tip over the sample (or the sample under the tip) in a raster pattern A feedback system to control the vertical position of the tip A sharp tip A way of sensing the vertical position of the tip A computer system that drives the scanner, measures data and converts the data into an image. A course positioning system to bring to tip into the general vicinity of the sample. Sample Figure 3-2 Atomic force microsc ope (after K.K. Harris 2000).

PAGE 63

46 Figure 3-3 TEM setup used to image atomic layers at the film/substrate interface. Electron Source Condenser Lens Specimen Objective Lens Back Focal Plane First Intermediate Image Intermediate Lens Second Intermediate Image Projector Lens Image

PAGE 64

47 Figure 3-4 SEM operation. Electron beam is rastered over the sample producing secondary electrons (a fter S.M. Donovan 1999). Cathode ray tube Screen Electron source Detector giving modulating signal Deflection coils DEFLECTION Secondary electron signal Sample

PAGE 65

48 Figure 3-5 Photoluminescence setup He-Cd Laser Sample mount Computer setup PMT Monochrometer Optical lens

PAGE 66

49 Figure 3-6 Ideal MOS I-V curve -1.5-1.0-0.50.00.51.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 flatband deep depletion (low frequency) (high frequency) inversion depletion accumulation Capacitance (arb. units)Voltage (arb. units) Figure 3-7 Ideal C-V curves of n-type MOS (after Johnson) I V

PAGE 67

50 Figure 3-8 Finished MOS capacitors design.

PAGE 68

51 A B Figure 3-9 Test Chamber for Radiation. A) Mounted samples in test chamber. B) Shutters of beam. BeamShutters

PAGE 69

52 Figure 3-10 Radiation Test Facili ty at Texas A&M University.

PAGE 70

53 CHAPTER 4 RADIATION AND THERMAL STABILITY OF VARIOUS TYPES OF GALLIUM NITRIDE Various types of GaN samples were test ed preand post-pr oton radiation using photoluminescence (PL). The effects of rapi d thermal annealing (RTA) were also investigated. The types of gallium nitride that were tested included n-type GaN, a high resistivity GaN (known as u-GaN), MBE grown GaN, and p-type GaN. PL was used to determine the initial luminescence state of the GaN and monitor changes in the luminescence profile after proton radiation and thermal annealing. RTA was used to determine if the radiation damage to the samp le could be corrected by heating the sample to various temperatures. 4.1 Characterization of MOCVD N-type GaN As grown commercial MOCVD GaN, s upplied by Uniroyal Optoelectronics (UOE), was probed using PL. Figure 4-1 show s the initial GaN luminescence profile at room temperature (300K) and low temper ature (15K). At 362nm(3.42eV), the primary transition or bandedge of the GaN is noted in addition to a defect band in the 450nm to 700nm range, which is related to various midga p defects in the gallium nitride. This yellow defect band gives n-GaN its distinctive luminescence color. Although the origin of this yellow band is heavily debated, th e defect origins are most likely grown-in nitrogen vacancies. Notice at 15K the la rge defect emission from 450nm to 700nm has been suppressed. According to Khanna et al., the suppression of the defect band at low

PAGE 71

54 temperatures is a result of reduced non-ra diative transitions and a lack of available phonons [52]. The UOE n-GaN samples were irradiated at two different ener gies of 10MeV and 40MeV with a total particle dose of 5 x109cm-2, which equals approximately 10 years of mission time in space applicati ons. Figures 4-2 and 4-3 show the post radiation damage to the GaN in comparison to the initial scan at 300K and 15K respectively. At 300K a significant decrease in luminescence intensity is noticed as well as a slight shift in the bandedge peak to 354nm. Post radiation at 15K shows an increase in the defect band of the profile which can be attributed to the increased amount of def ects within the bulk material that have now be come radiativ e centers for electron –hole recombination. The bandedge is significantly decreased also due to the introduction of non-radiative defects. Also due to the radiation damage the UOE n-GaN samples luminescenced a pale green color and not the customary yellow that char acterizes GaN. Based on previous work done by other groups, the reason for this decrease in intensity and change in color is an increase in lattice defects. After a year, the samples were rescanned with PL to determine the luminescence profile. Figure 4-4 shows a comparison to th e initial GaN scans and the post radiated GaN. After one year, the defect lumines cence returns to its pre-irradiated state. Similarly, there is a partial though not comple te recovery of the bandedge emission. This change in the profile from the immediat e post radiation scans implies either long life-time traps have relaxed or surface carrier loss has b een suppressed via oxidation or adsorption of adventitious species such as carbon. The samples still luminescence with a

PAGE 72

55 pale green-yellow and the PL profile was not completely returned to the original scan which shows that not all the changes cau sed by irradiation were corrected. In order to determine if the radiated sample s could be returned to their initial profile by thermal treatment, the UOE n-GaN samples were annealed using an RTA setup with varying temperatures ranging from 200oC to 900oC and held for 60 seconds at each temperature. After each temperature anneal, the GaN samples were rescanned using PL. In order to better monito r the changes in the n-Ga N induced by annealing, a nonirradiated n-GaN sample from the same wafer as the irradiated samples was also annealed and PL scanned. Figures 4-5 through 4-7 show s a comparison of the annealed radiated and non-radiated samples of n-GaN. After annealing at 200oC a significant increase in PL intensity was observed for all samples but at 300oC the intensity decreases well below the original level. It seems unlikely that diffusi on of nitrogen and gallium atoms to vacancies or interstitial sites within the GaN lattice has occurred at such a low temperature. Thus, it is possible that some surface phenomena are responsible, though the exact mechanism is as yet unknown. At higher temperatures of 400oC to 900oC, no large changes to the PL profiles in either intensity or bandedge shifts were obs erved. Based on figure 4-8, the GaN has reached a fairly steady configuration of defects within the lattice and that no more recovery or improvement of the samples can be achieved without completely damaging the GaN substrate. Figure 4-8 is a calculated trend of the annealed sample changes based on ratio comparisons of norma lized bandedge peaks of all annealed GaN to the GaN control peak intensity of 1.4812. After 600oC, there is a gradual decrease in bandedge intensity indicating th at further annealing is dama ging the optical properties of the GaN.

PAGE 73

56 After reviewing the results of the first UOE n-GaN, a further study of other types of GaN with different growth recipes and growth systems was undertaken in order to further investigate the phenomena at 200oC. Another UOE n-GaN sample from another wafer, which will be referred to as the 2nd UOE n-GaN sample, was scanned with PL, radiated at 10MeV and 40MeV and annealed from 200oC to 900oC to determine the 2nd UOE n-GaN trend. Figure 4-9 shows the pre and post radiation PL profiles of the 2nd UOE n-GaN. In comparison, both UOE n-GaN samples show the same basic trends in the radiation and thermal experiments. Figures 4-10 through 4-12 show the radiation and thermal effects on PL profiles for the 2nd UOE n-GaN sample. All of the data from the 2nd UOE n-GaN sample was collected immediately after radia tion. As noted before after annealing at 200oC a large increase in the PL intensity is observed. Thus far, this phenomenon has shown to be unique to all UOE GaN (n-GaN and u-GaN) samples and was not observed for any of the other GaN types, which will be discussed in later sections. This phenomenon could have various reasons, such as sample preparation, growth parameters for UOE GaN, surface effects or material doping, which are unknown and proprietary information at the present time. In figure 4-13, the normalized bandedge peak intensity versus annealing temperature is shown for the 2nd UOE n-GaN sample. The trend in the plot generally follows the first UOE n-GaN w ith no significant changes in the profiles after 400oC. For both samples, annealing was not effective in completely restoring the spectra to the pre-irradiated prof ile, and after an nealing at 900oC the luminescence ratio is almost equal to that of the radiated sample ratio. At the higher temperatures, it is likely that nitrogen is being driven out of the sample and leavi ng behind nitrogen defects that essentially creates nonradiative centers.

PAGE 74

57 In addition, n-GaN grown by an MOCVD system at the University of Florida (UF) was also studied and underwent the same e xperiments as the UOE n-GaN to explore the potential variability in MOCVD GaN. The Ga N grown at UF sample will be referred to as UF n-GaN through out the paper. Figure 4-14 shows the pre and post radiation PL scans, like the UOE n-GaN, irradiation of the UF n-GaN produced a significant drop in PL intensity. Figures 4-15 through 4-17 shows the results of rapid thermal annealing from 200oC to 900oC for each UF n-GaN sample. Figure 4-18 plots the bandedge peaks normalized to 25.069 in intensity. Unlike the UOE n-GaN, the non-irradiated UF n-GaN shows a significant drop in both bandedge and defect lu minescence for all annealing temperatures. Annealing of the irradiated ma terial produces little to no improvement in the intensity of the PL. 4.2 Characterization of MOCVD U-type GaN As with the n-GaN samples, a set of UOE verses UF u-GaN was tested and compared. Generally u-GaN has lower lumines cence intensity due to a reduction in free carriers. Figure 4-19 shows the changes in th e PL scan after proton radiation at 10MeV and 40MeV. Notice that there is a severe reduc tion in the already low peak intensity. The broadness of the bandedge peak indicates the pr esence of defects in the gap of the GaN. Figures 4-20 through 4-22 depicts the effects of thermal annealing of the samples. In each of the samples there is an increase at 200oC, which, as mentioned before in previous section, is mostly likely surface effects of the GaN since the intensity continuously drops with increasing annealing temperatures beyond the temperature. Normalizing the bandedges peak intensity to the control sample bandedge shows the trend of increasing and decreasing intensity with anneal ing temperature (figure 4-23). At 400oC there is a severe decrease in intensity and no additional dr astic change or recovery of the samples.

PAGE 75

58 The UF u-GaN shows higher peak inte nsity than the UOE u-GaN possibly indicating a better quality sample. Figure 4-24 shows the pre and post radiation scans of the UF u-GaN and in figures 4-25 through 4-27 the thermal annealing is shown. The main trend of both the UOE u-GaN samples and the UF u-GaN is that at 400oC the control samples and the radiated samples are about the same in peak intensity and continue in a linear fashion to 900oC. Figure 4-28 exhibits this tre nd in the UF u-GaN. Radiation reduced the bandedge but only slightly reduced the defect luminescence. As with the UF n-GaN, annealing at all temperat ures reduce the PL intensity. 4.3 Characterization of MBE Grown Ga-Polar GaN MBE gallium nitride is known to be highly resistive mostly likely due to selfcompensation of the Ga and exhibit for weak PL emissions due to lack of free carriers available in the samples for emission tran sitions. Figure 4-29 show s the comparison of the pre and post radiation scan s of the Ga-polar GaN sample. The bandedge peak after irradiation is no longer visibl e in the PL scan mostly in dicating a large increase in optically active defects in the samples or de gradation of the surface. After annealing each sample (figures 4-30 through 4-32), there was a s light increase in the p eak intensity at the higher temperatures but it neve r recovers to the original peak intensity. Figure 4-33 shows the normalized bandedge peak intensit ies to the control versus the annealing temperatures. 4.4 Characterization of MOCVD P-type GaN The comparison of UF MOCVD grown p-type GaN was studied with PL, annealing, and Hall to determine the behavior of the sample after proton irradiation and thermal testing. The samples were taken from the same wafer but one half of the wafer received a post activation anneal of the Mg implantation at 750oC for 30 seconds in the

PAGE 76

59 RTA. The as grown p-GaN will be referred to as P-GaN and the part of the wafer that received the post growth activa tion anneal will be referred to as Activated P-GaN. In figure 4-34 is the PL comparison of each control sample. Figures 4-35 and 4-36 shows the pre and post PL scans of the p-GaN and ac tivated p-GaN samples. Note the difference in peak intensity of the control samples and th en the observed decrease in intensity of the radiated p-GaN samples but the increase in the activated p-GaN samples. It is possible that the radiation in the activated sample s has caused ionization of the Mg acceptors. Figures 4-37 through 4-38 show the PL s cans after annealing each of the p-GaN (unactivated and activated) control samples and figures 4-39 and 4-42 captures the trend of the 10MeV and 40MeV radiated p-GaN (un-activat ed and activated) samples. Each of the samples experienced a sharp decline in intensity at 400oC and again at 600oC. They each increased in peak intensity at 900oC which mostly likely can be attributed to defect annealing. Each of the un-activated p-GaN sa mples generally follows the same trend with or without radiation, whereas th e activated p-GaN shows a larger spread in th e ratios of the normalized bandedge peaks (figures 443 and 4-44). Figure 4-45 compares the two plots of bandedge peaks versus annealing te mperature where it exhibits the difference in the changes of the samples. The as grow n p-GaN look much more linear than the activated p-GaN. 4.5 Summary Based on the data collected, some general tr ends for the GaN can be noted. First is that radiation of the sample s decrease the PL luminescence intensity. Second, an increase of PL intensity at 200oC has shown to be unique to the UOE GaNs and as-grown P-GaN. After approximately 400oC, the drastic decline in peak intensity stopped and became less

PAGE 77

60 significant in change with higher temperat ures. These defects become non-radiative centers and trap carriers that contribute to optical transiti ons in the bandgap of the GaN.

PAGE 78

61 300400500600700800 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Peak IntensityWavelen g th 1st UOE n-GaN Control @ 300K 1st UOE n-GaN Control @ 15K Figure 4-1 Initial PL scans of non-radiated 1st UOE n-GaN at temperatures 300K and 15K.

PAGE 79

62 300400500600700800 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 1st UOE n-GaN Control Rad 10MeV Rad 40MeVPeak IntensityWavelength Figure 4-2 Room Temperature PL analysis of 1st UOE n-GaN before and after radiation.

PAGE 80

63 300400500600700800 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 1st UOE n-GaN Control Rad 10MeV Rad 40MeVPeak IntensityWavelen g th Figure 4-3 15K PL analysis of 1st UOE n-GaN before and after radiation.

PAGE 81

64 300400500600700800 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 1st UOE n-GaN Control Radiated 10MeV Radiated 40MeV 10MeV 1 year later 40MeV 1 year laterPeak IntensityWavelength Figure 4-4 Comparison of 300K 1st UOE n-GaN PL scans, pre-radiation, immediately post radiation, and post radiation 1 year later.

PAGE 82

65 A 300400500600700800 0.0 0.5 1.0 1.5 Peak IntensityWavelength 1st UOE n-GaN Control Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C B350360370380 0.0 0.5 1.0 1.5 Peak IntensityWavelength 1st UOE n-GaN Control Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 4-5 PL analysis of annealed un-radi ated 1st UOE n-GaN. A) PL Spectra B) Bandedge inset.

PAGE 83

66 A300400500600700800 0.0 0.7 1.4 2.1 Peak IntensityWavelen g th 1st UOE n-GaN Control Radiated 10MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C B350360370380 0.0 0.7 1.4 2.1 Peak IntensityWavelength 1st UOE n-GaN Control Radiated 10MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 4-6 PL analysis of annealed 10MeV radiated 1st UOE n-GaN A) PL Spectra B) Bandedge inset (below).

PAGE 84

67 A 300400500600700800 0.0 0.5 1.0 1.5 Peak IntensityWavelen g th 1st UOE n-GaN Control Radiated 40MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C B350360370380 0.0 0.5 1.0 1.5 Peak IntensityWavelength 1st UOE n-GaN Control Radiated 40MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 4-7 PL analysis of annealed 40MeV ra diated 1st UOE n-GaN. A) PL Spectra B) Bandedge inset.

PAGE 85

68 02004006008001000 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 Peak Intensity RatioTemperature 1st UOE n-GaN Control Radiated 10MeV Radiated 40MeV Figure 4-8 Normalized bandedge peaks of all annealed 1st UOE GaN to the 1st UOE GaN control peak intensity of 1.4812 vs. the annealing temperatures. 0 temperature indicates the initial state of the GaN.

PAGE 86

69 300400500600700800 0 2 4 6 8 10 12 Peak IntensityWavelength 2nd UOE n-GaN Control Radiated 10MeV Radiated 40MeV Figure 4-9 Pre and post radiation PL analysis of 2nd UOE n-GaN.

PAGE 87

70 A300400500600700800 0 5 10 15 Peak Intensit y Wavelength 2nd UOE n-GaN Control Annealed 200C Annealed 400C Annealed 600C Annealed 800C Annealed 900C B350360370380 0 5 10 Peak IntensityWavelength 2nd UOE n-GaN Control Annealed 200C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 4-10 PL analysis of un-radiated a nnealed 2nd UOE n-GaN. A) PL Spectra B) Bandedge inset.

PAGE 88

71 A300400500600700800 0 2 4 6 8 10 12 Peak IntensityWavelength 2nd UOE n-GaN Control Radiated 10MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C B350360370380 0 4 8 Peak IntensityWavelength 2nd UOE n-GaN Control Radiated 10MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 4-11 PL analysis of annealed 10MeV radiated UOE n-GaN. A) PL Spectra B) Bandedge inset

PAGE 89

72 A300400500600700800 0 4 8 12 Peak IntensityWavelength 2nd UOE n-GaN Control Radiated 40MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C B350360370380 0 4 8 Peak IntensityWavelength 2nd UOE n-GaN Control Radiated 40MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 4-12 PL analysis of annealed 40MeV radiated UOE n-GaN. A) PL Spectra B) Bandedge inset

PAGE 90

73 02004006008001000 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Bandedge RatiosTemperature 2nd UOE N-GaN Control Rad 10MeV Rad 40MeV Figure 4-13 Normalized bande dge peaks of annealed 2nd UOE n-GaN to the 2nd UOE nGaN control peak intensity of 12.354 vs. the annealing temperatures.

PAGE 91

74 300400500600700800 0 5 10 15 20 25 Peak IntensityWavelength UF n-GaN Control Radiated 10MeV Radiated 40MeV Figure 4-14 Pre and post radiation PL analysis of UF n-GaN.

PAGE 92

75 A300400500600700800 0 9 18 27 Peak IntensityWavelength UF n-GaN Control Annealed 200C Annealed 400C Annealed 600C Annealed 800C Annealed 900C B360370 0 9 18 Peak IntensityWavelength UF n-GaN Control Annealed 200C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 4-15 PL analysis of annealed un-radiat ed UF n-GaN. A) PL Spectra B) Bandedge inset.

PAGE 93

76 A300400500600700800 0 9 18 27 Peak IntensityWavelength UF n-GaN Control Radiated 10MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C B350360370380 0 9 18 Peak IntensityWavelength UF n-GaN Control Radiated 10MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 4-16 PL analysis of annealed 10MeV radiated UF n-GaN. A) PL Spectra B) Bandedge inset

PAGE 94

77 A300400500600700800 0 9 18 27 Peak IntensityWavelen g th UF n-GaN Control Radiated 40MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C B340350360370380 0 9 18 Peak IntensityWavelength UF n-GaN Control Radiated 40MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 4-17 PL analysis of annealed 40MeV radiated UF n-GaN. A) PL Spectra B) Bandedge inset

PAGE 95

78 02004006008001000 0.0 0.2 0.4 0.6 0.8 1.0 Bandedge RatiosTemperature UF N-GaN Control Rad 10MeV Rad 40MeV Figure 4-18 Normalized bandedge peaks of anne aled UF n-GaN to the UF n-GaN control peak intensity of 25.069 vs. the annealing temperatures.

PAGE 96

79 300400500600700800 0.00 0.02 0.04 0.06 0.08 0.10 0.12 Peak IntensityWavelength UOE u-GaN Control Radiated 10MeV Radiated 40MeV Figure 4-19 Pre and Post Ra diation of UOE u-GaN.

PAGE 97

80 A300400500600700800 0.00 0.02 0.04 0.06 0.08 0.10 0.12 Peak IntensityWavelength UOE u-GaN Control Annealed 200C Annealed 400C Annealed 600C Annealed 800C Annealed 900C B350360370 0.00 0.04 0.08 Peak IntensityWavelength UOE u-GaN Control Annealed 200C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 4-20 PL analysis of annealed un-ra diated UOE u-GaN. A) PL Spectra B) Bandedge inset.

PAGE 98

81 A300400500600700800 0.00 0.02 0.04 0.06 0.08 0.10 0.12 P ea k I n t ens it yWavelength Control UOE u-GaN Radiated 10MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C B350360370380 0.00 0.04 0.08 Peak IntensityWavelength Control UOE u-GaN Radiated 10MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 4-21 PL analysis of annealed 10MeV radiated UOE u-GaN. A) PL Spectra B) Bandedge inset.

PAGE 99

82 A300400500600700800 0.00 0.02 0.04 0.06 0.08 0.10 0.12 P ea k I n t ens it yWavelength Control UOE u-GaN Radiated 40MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C B340350360370380 0.00 0.04 0.08 0.12 Peak IntensityWavelength Control UOE u-GaN Radiated 40MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 4-22 PL analysis of annealed 40MeV radiated UOE u-GaN. A) PL spectra B) Bandedge inset.

PAGE 100

83 02004006008001000 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Bandedge RatiosTemperature UOE U-GaN Control Rad 10MeV Rad 40MeV Figure 4-23 Normalized bandedge peaks of annealed UOE u-GaN to the UOE u-GaN control peak intensity of 0.1162 vs. the annealing temperatures.

PAGE 101

84 300400500600700800 0.0 0.3 0.6 0.9 1.2 1.5 Peak IntensityWavelength UF U-GaN Control Rad 10MeV Rad 40MeV Figure 4-24 Pre and post radiati on PL analysis of UF u-GaN

PAGE 102

85 A300400500600700800 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Peak IntensityWavelength UF U-GaN Control Annealed 200C Annealed 400C Annealed 600C Annealed 800C Annealed 900C B340360380 0.0 0.5 1.0 Peak IntensityWavelength UF U-GaN Control Annealed 200C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 4-25 PL analysis of annealed un-radiat ed UF u-GaN. A) PL Spectra B) Bandedge inset.

PAGE 103

86 A300400500600700800 0.0 0.5 1.0 1.5 Peak IntensityWavelength UF U-GaN Control Rad 10MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C B340360380 0.0 0.5 1.0 Peak IntensityWavelength UF u-GaN Control Rad 10MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 4-26 PL analysis of annealed 10MeV radiated UF u-GaN. A) PL Spectra B) Bandedge inset.

PAGE 104

87 A300400500600700800 0.0 0.5 1.0 1.5 Peak IntensityWavelength UF U-GaN Control Rad 40MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C B340360380 0.0 0.5 1.0 Peak IntensityWavelength UF U-GaN Control Rad 40MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 4-27 PL analysis of annealed 40MeV radiated UF u-GaN. A) PL Spectra B) Bandedge inset.

PAGE 105

88 02004006008001000 0.0 0.2 0.4 0.6 0.8 1.0 Bandedge RatiosTemperature UF U-GaN Control Rad 10MeV Rad 40MeV Figure 4-28 Normalized bandedge peaks of anne aled UF u-GaN to the UF u-GaN control peak intensity of 1.2997 vs. the annealing temperatures.

PAGE 106

89 300400500600700800 0.00 0.05 0.10 0.15 0.20 0.25 Peak IntensityWavelength MBE GaN Control Radiated 10MeV Radiated 40MeV Figure 4-29 Pre and Post Radiation PL analysis of MBE Ga-polar GaN.

PAGE 107

90 A300400500600700800 0.00 0.05 0.10 0.15 0.20 0.25 Peak IntensityWavelength MBE GaN Control Annealed 200C Annealed 400C Annealed 600C Annealed 800C Annealed 900C B350360370380 0.006 0.012 Peak IntensityWavelength MBE GaN Control Annealed 200C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 4-30 PL analysis of annealed un-radiat ed MBE Ga-polar GaN. A) PL Spectra B) Bandedge inset.

PAGE 108

91 A300400500600700800 0.00 0.05 0.10 0.15 0.20 0.25 Peak IntensityWavelength MBE GaN Control Rad 10MeV Anneal 200C Anneal 400C Anneal 600C Anneal 800C Anneal 900C B350360370380 0.000 0.005 0.010 0.015 Peak IntensityWavelength MBE GaN Control Rad 10MeV Anneal 200C Anneal 400C Anneal 600C Anneal 800C Anneal 900C Figure 4-31 PL analysis of annealed 10MeV radiated MBE GaN. A) PL Spectra B) Bandedge inset.

PAGE 109

92 A300400500600700800 0.00 0.05 0.10 0.15 0.20 0.25 Peak IntensityWavelength MBE GaN Control Rad 40MeV Anneal 200C Anneal 400C Anneal 600C Anneal 800C Anneal 900C B350360370380 0.000 0.006 0.012 Peak IntensityWavelength MBE GaN Control Rad 40MeV Anneal 200C Anneal 400C Anneal 600C Anneal 800C Anneal 900C Figure 4-32 PL analysis of annealed 40MeV radiated MBE GaN. A) PL Spectra B) Bandedge inset.

PAGE 110

93 02004006008001000 0.4 0.6 0.8 1.0 1.2 1.4 Bandedge RatiosTemperature MBE Ga-polar GaN Control MBE Ga-polar GaN Rad 10MeV MBE Ga-polar GaN Rad 40MeV Figure 4-33 Normalized bandedge peaks of annealed MBE Ga-polar GaN to the MBE Ga-polar GaN control peak intensity of .01016 vs. the annealing temperatures.

PAGE 111

94 300400500600700800 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Peak Intensit y Wavelength P-GaN Control RTA P-GaN Control Figure 4-34 Comparison of PL scans of as grown P-GaN and Activated P-GaN.

PAGE 112

95 300400500600700800 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Peak IntensityWavelength P-GaN Control Rad 10MeV Rad 40MeV Figure 4-35 Pre and post radiation PL analysis of as-grown P-GaN.

PAGE 113

96 300400500600700800 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 Peak IntensityWavelength Activated P-GaN Control Rad 10MeV Rad 40MeV Figure 4-36 Pre and post radiation PL analysis of Activated P-GaN.

PAGE 114

97 300400500600700800 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Peak IntensityWavelength P-GaN Control Annealed 200C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 4-37 PL analysis of annealed as-grown un-radiated P-GaN.

PAGE 115

98 300400500600700800 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 Peak IntensityWavelength Activated P-GaN Control Annealed 200C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 4-38 PL analysis of anneal ed un-radiated Activated P-GaN.

PAGE 116

99 300400500600700800 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Peak IntensityWavelength P-GaN Control Rad 10MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 4-39 PL analysis of a nnealed 10MeV as-grown P-GaN

PAGE 117

100 300400500600700800 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Peak IntensityWavelength P-GaN Control Rad 40MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 4-40 PL analysis of ann ealed 40MeV as-grown P-GaN.

PAGE 118

101 300400500600700800 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 Peak IntensityWavelength Activated P-GaN Control Rad 10MeV Annealed 200C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 4-41 PL analysis of annealed Activated P-GaN radiated at 10MeV.

PAGE 119

102 300400500600700800 -0.1 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.7 Peak IntensityWavelength Activated P-GaN Control Rad 40MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 4-42 PL analysis of annealed Activated P-GaN radiated at 40MeV.

PAGE 120

103 02004006008001000 0.0 0.2 0.4 0.6 0.8 1.0 Bandedge RatiosTemperature P-GaN Control P-GaN Rad 10MeV P-GaN Rad 40MeV Figure 4-43 Normalized bandedge peaks of anne aled as grown P-GaN to the as grown PGaN control peak intensity of 2.7808 vs. the annealing temperatures.

PAGE 121

104 02004006008001000 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0 3.2 3.4 3.6 3.8 4.0 Bandedge RatiosTemperature Activated P-GaN Control Activated P-GaN Rad 10MeV Activated P-GaN Rad 40MeV Figure 4-44 Normalized bandedge peaks of a nnealed Activated P-GaN to the Activated P-GaN control peak intensity of .42224 vs. the annealing temperatures.

PAGE 122

105 02004006008001000 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0 3.2 3.4 3.6 3.8 4.0 Bandedge RatiosTemperature P-GaN Control P-GaN Rad 10MeV P-GaN Rad 40MeV Activated P-GaN Control Activated P-GaN Rad 10MeV Activated P-GaN Rad 40MeV Figure 4-45 All Normalized P-GaN bandedge peaks to control peaks of 2.7808 and .42224 respectively.

PAGE 123

106 CHAPTER 5 RADIATION AND THERMAL STABILIT Y OF NOVEL OXIDES ON GALLIUM NITRIDE Dielectrics on GaN were tested prea nd post-proton radiation using, XRR and PL. The effects of rapid thermal annealing (RTA) were also investigated. PL was used to determine the initial luminescence state of the GaN. GaN has a distinct luminescence profile, changes in which can be monitored by observing the peak intensity changes and shifts in the peak wavelengths. XRR was used to monitor the morphological changes between the interface of the dielectric and the GaN. Observations of the periodic oscillations and amplitude of the oscillatio ns can show roughing at the interface due to particle bombardment by the radiation. Annealin g was used to determine if the damage to the sample could be corrected by heating th e sample up to various temperatures. After analyzing the results of the GaN sample s, three different oxides of ~400 Sc2O3, ~200 MgO with a ~200 Sc2O3 capping layer, and ~200 MgCaO with a ~200 capping layer were epitaxially grown on UOE n-GaN to determine if the oxide layers would protect the interface between th e oxide and the GaN as well as reduce the bulk damage to the GaN. The oxide/GaN samples were initia lly characterized with PL and XRR. The samples were then irradiated with protons at two different en ergies of 10MeV and 40MeV with a total dose of 5 x 109cm-2 and were then re-characterized using the same techniques.

PAGE 124

107 5.1 Characterization of Sc2O3 on GaN Figures 5-1 shows the initial PL and XRR profiles of the Sc2O3/GaN. The starting roughness at the interface of the oxide and se miconductor and the surface of the oxide are ?? and ?? respectively. In figure 5-2 a noticeable decrease in bandedge intensity is observed for both radiation energies for the Sc2O3/GaN sample but an increase in the defect density intensity for the sample irra diated at 40MeV. The PL scan taken at 15K shows an increase in the defect band emi ssion and a large bandedge intensity increase. The XRR data in figure 5-3 s hows the reduction in periodicit y and amplitude of the XRR scan relative to the initial sca n, which indicates roughening of the Sc2O3/GaN sample interface, due to particle bombardment of the proton radiation. The oxide sample exhibited the same pale green color as the non-oxide/GaN. The samples were then scanned a year la ter to determine any changes in the PL profiles. Figure 5-4 shows a comparison of th e initial, immediate po st radiation, and post radiation one year later. As with the non-oxide GaN, thes e samples also showed that some relaxation had occurred within the sample s. Notice the return of the peak intensity in comparison to the immediate post radia tion scans, however, the sample did not completely return to its initial as grow n state. Also, the oxide/GaN samples still luminescenced a pale green-yellow color indi cating that the samples had not completely recovered from the damage received from the irradiation. The samples were subjected to annealing to determine weather any more recovery of the damage could be seen. A companion nonradiated sample was subjected to the same annealing temperature ranges of 200oC to 900oC in order to compare them to the irradiated annealed samples to determine how much recovery or damage is done by just annealing alone. Figures 5-5 thr ough 5-7 show the PL scans Sc2O3/GaN samples after

PAGE 125

108 annealing from 200oC to 900oC. Notice the large increase in the scans at 200oC and then a decrease at 300oC. The Sc2O3/GaN PL scans show an increase from 200oC to 400oC in both bandedge intensity and defect density intensity. The increases appear to be approximately the same for each annealed te mperature, which can possibly indicate that annealing is producing more defects in the sa mple. This phenomenon was also seen in the non-oxide GaN, which indicates that this is possibly a surface effect of the GaN. After annealing to 400oC, the samples were rescanned usi ng XRR to look at the effects the annealing had on the interfaces. Figures 5-8 a nd 5-9 shows the comparison of the initial XRR scans to the post-radiated post-anneal ed scans. Annealing the samples to 400oC has shown further roughening of the interfaces of the oxide to the GaN substrate. After annealing the samples to 400oC, they were then scanned by XRR to look at the effects of just annealing on the interf aces of the samples. XRR showed a loss of periodic oscillations as shown in figure 57 and as with the radiated samples, the interfaces have roughened with annealing. Results of roughening are current leakage via interface state traps once the oxide/GaN is fabr icated into a MOS device, which shortens the gate breakdown voltage of the device. 5.2 Characterization of Sc2O3/MgO on GaN Figures 5-11 and 5-12 shows the in itial PL and XRR profiles of the Sc2O3/MgO/GaN. For the Sc2O3/MgO samples, both showed a decrease in bandedge and defect density intensity at both radiation energi es at room temperatur e but at 15K there is a marked increase in the intensity of the bandedge and defect densities. Figure 5-13 shows the damage to the interface via loss of oscillations at 40MeV. The MgCaO/ Sc2O3 sample displayed a decrease in bandedge a nd defect density at bot h energies at room temperature and low temperature of 15K as shown by figure 4-16. For the MgO/ Sc2O3,

PAGE 126

109 the PL scans are virtually identical indicating that there is no signifi cant change has taken place to change the luminescence profile. Ho wever, the XRR scan does show roughening of one or more of th e samples has occurred. 5.3 Characterization of Sc2O3/MgCaO on GaN Figures 5-1 and 5-2 show the ini tial PL and XRR profiles of the Sc2O3/MgCaO/GaN. In comparison to the other oxi de samples the PL scan shows a slight difference in the defect density peaks, noted by the more narrowed spikes in the 450nm to 700nm range. A possible reason for this is th e lattice matching between the MgCaO/GaN interface. This is also evident in the XRR co mparison of the pre and post irradiation data in figure 4-17. The XRR profile has some change at the interface but not as much as with the other less latticed matched oxides. In figure 4-30 the post annealed PL scans of MgCaO/ Sc2O3 show identical scan at 200oC and 300oC but reveals an increase in both bandedge and defect densities at 400oC. This is most likely caused by the creation of defects and traps since both peaks have increased about the same. The XRR scan show s a decrease in the oscillations, which as stated before indicated roughening at one or more the interfaces. 5.4 Summary The roughening of the oxide/GaN interface s uggests that operating temperatures of any device fabricated from the MgO and MgCaO oxides maybe be limited. As with the non-oxide GaN samples the irradiation of the oxide samples still significantly decreases the bandedge emission and increases the de fect emission. The radiation also causes roughening of the oxide/semiconductor interface and roughening at the surface. After one year, the samples show some relaxation of long life-time traps that degrade the PL emission, as was shown with the non-oxide Ga N. The effect of annealing on the oxide

PAGE 127

110 samples also causes roughening at the su rface and interface in the MgO and MgCaO samples. Of the three different oxides, Sc2O3 shows the least amount of change after annealing to 400oC. The MgO with capping layer shows the most change based on the XRR scans that was done implying that although the capping layer has slowed degradation of the MgO, there is still some hydrolyzing of the MgO occurring. Radiation of the samples makes them more susceptible to annealing damage because of the defects already present. As stated in earlier sections the purpose of the altern ate dielectrics is to find one that allowed for elev ated operating temperatures. 300400500600700800 0.0 0.5 1.0 1.5 2.0 Pre Rad @ 300K Pre Rad @ 15K Peak IntensityWavelength

PAGE 128

111 123 1 10 100 1000 10000 100000 Pre Rad Sc2O3XRR Counts Figure 5-1 Preliminary Sc2O3/GaN A) PL spectra and B) XRR scan.

PAGE 129

112 A300400500600700800 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Pre Rad @ 300K Post Rad 10MeV @ 300K Post Rad 40MeV @ 300KPeak IntensityWavelength B300400500600700800 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Pre Rad @ 15K Post Rad 10MeV @ 15K Post Rad 40MeV @ 15KPeak IntensityWavelength Figure 5-2 Sc2O3/GaN PL analysis. A) PL at 300K. B) PL at 15K.

PAGE 130

113 123 1 10 100 1000 10000 100000 Pre Rad Post 10MeV Rad Post 40MeV RadXRR Counts Figure 5-3 Sc2O3/GaN XRR scans showing the change in oscillations.

PAGE 131

114 A300400500600700800 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Pre Rad Post 10MeV Rad Post 40MeV Rad Post 10MeV Rad 1 year later Post 40MeV Rad 1 year laterPeak IntensityWavelength B340360380 0.0 0.2 0.4 0.6 Pre Rad Post 10MeV Rad Post 40MeV Rad Post 10MeV Rad 1 year later Post 40MeV Rad 1 year laterPeak IntensityWavelength Figure 5-4 Sc2O3/GaN PL scans of pre-radiation, immediate post radiation and post irradiation 1 year later. A) PL spectra B) Bandedge inset.

PAGE 132

115 A300400500600700800 0.0 0.2 0.4 0.6 0.8 Peak IntensityWavelength ScO/GaN Control Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C B340360380 0.0 0.2 0.4 Peak IntensityWavelength ScO/GaN Control Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 5-5 PL scans of samples annealed non-radiated Sc2O3/GaN to 900C. A) PL Spectra B) Bandedge inset

PAGE 133

116 A300400500600700800 0.0 0.5 1.0 1.5 Peak IntensityWavelength ScO/GaN Control Rad 10MeV 1 year later Rad 10MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C B360370380 0.0 0.5 1.0 Peak IntensityWavelength ScO/GaN Control Rad 10MeV 1 year later Rad 10MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 5-6 Sc2O3/GaN post radiation anneals to 900oC for radiation energies of 10MeV. A) PL spectra B) Bandedge inset.

PAGE 134

117 A300400500600700800 0.0 0.4 0.8 1.2 Peak IntensityWavelength ScO/GaN Control Rad 40MeV 1 year later Rad 40MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C B350360370380390 0.0 0.4 0.8 1.2 Peak Intensit y Wavelength ScO/GaN Control Rad 40MeV 1 year later Rad 40MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 5-7 Sc2O3/GaN post radiation anneals to 900oC for radiation energies of 40MeV. A) PL Spectra B) Bandedge inset.

PAGE 135

118 123 1 10 100 1000 10000 100000 Control annealed 400CXRR Counts Figure 5-8 Sc2O3/GaN XRR scans of unradiate d annealed sample at 400C.

PAGE 136

119 123 1 10 100 1000 10000 100000 1000000 Pre Rad Post 10MeV Rad Post 40MeV Rad Post 10MeV Rad Annealed to 400C Post 40Mev Rad Annealed to 400CXRR Counts Figure 5-9 Sc2O3/GaN XRR scans pre and post radiation annealed at 400oC.

PAGE 137

120 02004006008001000 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Normalized Bandedge RatioTemperature Sc2O3/GaN Control Radiated 10MeV Radiated 40MeV Rad 10MeV 1Yr. Later Rad 40MeV 1Yr. Later Figure 5-10 Normalized bande dge peaks of annealed Sc2O3/GaN to the Sc2O3/GaN control peak intens ity of .5139 vs. annealing temperature.

PAGE 138

121 A300400500600700800 0.0 0.2 0.4 0.6 0.8 1.0 Peak IntensityWavelength Pre Rad @ 300K Pre Rad @ 15K B123 1 10 100 1000 10000 100000 Pre Rad ScO/MgO/GaNXRR Counts Figure 5-11 Preliminary Sc2O3(20nm)/MgO(20nm)/GaN PL spectra and XRR. A) PL spectra B) XRR scan.

PAGE 139

122 A300400500600700800 0.0 0.2 0.4 0.6 0.8 1.0 Pre Rad @ 300K Post 10MeV Rad @ 300K Post 40MeV Rad @ 300KPeak IntensityWavelength B300400500600700800 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Pre Rad @ 15K Post 10MeV Rad @ 15K Post 40MeV Rad @ 15KPeak IntensityWavelength Figure 5-12 Sc2O3(20nm)/MgO(20nm)/GaN PL scans. A) PL at 300K B) PL at 15K.

PAGE 140

123 123 1 10 100 1000 10000 100000 Pre Rad Post 40MeV RadXRR Counts Figure 5-13 Sc2O3(20nm)/MgO(20nm)/GaN XRR scan showing the change in oscillations at 40 MeV.

PAGE 141

124 A300400500600700800 0.0 0.2 0.4 0.6 0.8 1.0 Pre Rad Post 10MeV Rad Post 40MeV Rad Post 10MeV Rad 1 year later Post 40MeV Rad 1 year laterPeak IntensityWavelength B350360370380 0.0 0.2 0.4 Pre Rad Post 10MeV Rad Post 40MeV Rad Post 10MeV Rad 1 year later Post 40MeV Rad 1 year laterPeak IntensityWavelength Figure 5-14 Sc2O3(20nm)/MgO(20nm)/GaN PL scans of pre-radiation, immediate post radiation and post irradiati on 1 year later. A) PL spectra B) Bandedge inset.

PAGE 142

125 A300400500600700800 0.0 0.2 0.4 0.6 0.8 1.0 Peak IntensityWavelength ScO/MgO/GaN Control Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C B350360370380 0.0 0.2 Peak IntensityWavelength ScO/MgO/GaN Control Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 5-15 PL scans of sample s annealed non-radiated Sc2O3(20nm)/MgO(20nm)/GaN to 900C. A) Pl Spectra B) Bandedge inset.

PAGE 143

126 A300400500600700800 0.0 0.4 0.8 1.2 Peak IntensityWavelength ScO/MgO/GaN Control Rad 10MeV 1 year later Rad 10MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C B360370380 0.0 0.4 0.8 Peak IntensityWavelength ScO/MgO/GaN Control Rad 10MeV 1 year later Rad 10MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 5-16 Sc2O3(20nm)/MgO(20nm)/GaN post radiation anneals to 900oC for radiation energies of 10MeV. A) PL spectra B) Bandedge inset.

PAGE 144

127 A300400500600700800 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Peak IntensityWavelength ScO/MgO/GaN Control Rad 40MeV 1 year later Rad 40MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C B360370380390 0.0 0.4 0.8 Peak IntensityWavelength ScO/MgO/GaN Control Rad 40MeV 1 year later Rad 40MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 5-17 Sc2O3(20nm)/MgO(20nm)/GaN post radiation anneals to 900oC for radiation energies of 40MeV. A) PL spectra B) Bandedge inset.

PAGE 145

128 123 0.1 1 10 100 1000 10000 100000 1000000 1E7 orginal as grown annealed at 400C orginal as grown 1 year laterXRR Counts Figure 5-18 Sc2O3(20nm)/MgO(20nm)/GaN XRR scan of annealed sample.

PAGE 146

129 123 1 10 100 1000 10000 100000 Pre Rad Post Rad 40MeV Post Rad 10MEV Annealed to 400 C Post Rad 40MeV Annealed to 400CXRR Counts Figure 5-19 Sc2O3(20nm)/MgO(20nm)/GaN XRR scans pre and post radiation annealed at 400oC.

PAGE 147

130 02004006008001000 0 1 2 3 4 5 6 Normalized Bandedge RatioTemperature Sc2O3/MgO/GaN Control Radiated 10MeV Radiated 40MeV Rad 10MeV 1Yr. Later Rad 40MeV 1Yr. Later Figure 5-20 Normalized bandedge peaks of annealed Sc2O3(20nm)/MgO(20nm)/GaN to the Sc2O3(20nm)/MgO(20nm)/GaN contro l peak intensity of .21556 vs. annealing temperature.

PAGE 148

131 A300400500600700800 0 1 2 3 4 5 6 Peak IntensityWavelength Pre Rad @ 300K Pre Rad @ 15K B123 1 10 100 1000 10000 100000 Pre Rad ScO/MgCaO/GaNXRR Counts Figure 5-21 Preliminary Sc2O3(20nm)/MgCaO(20nm)/GaN. A) PL spectra B) XRR scan.

PAGE 149

132 A300400500600700800 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 Pre Rad @ 300K Post 10MeV Rad @ 300K Post 40MeV Rad @ 300KPeak IntensityWavelength B300400500600700800 0 1 2 3 4 5 6 Pre Rad @ 15K Post 10MeV Rad @ 15K Post 40MeV Rad @ 15KPeak IntensityWavelen g th Figure 5-22 Sc2O3(20nm)/MgCaO(20nm)/GaN PL scans. A) PL at 300K B) PL at 15K (bottom).

PAGE 150

133 123 1 10 100 1000 10000 100000 Pre Rad Post 10MeV Rad Post 40MeV RadXRR Counts Figure 5-23 Sc2O3(20nm)/MgCaO(20nm)/GaN XRR scan showing the change in oscillations.

PAGE 151

134 A300400500600700800 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 Pre Rad Post 10MeV Rad Post 40MeV Rad Post 10MeV Rad 1 year later Post 40MeV Rad 1 year laterPeak IntensityWavelength B360380 0.0 0.4 0.8 1.2 Pre Rad Post 10MeV Rad Post 40MeV Rad Post 10MeV Rad 1 year later Post 40MeV Rad 1 year laterPeak IntensityWavelen g th Figure 5-24 Sc2O3(20nm)/MgCaO(20nm)/GaN PL scan s of pre-radiation, immediate post radiation and post irra diation 1 year later. A) PL spectra B) Bandedge inset.

PAGE 152

135 A300400500600700800 0.0 0.4 0.8 1.2 1.6 Peak IntensityWavelength ScO/MgCaO/GaN Control Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C B340360380 0.0 0.4 0.8 1.2 Peak IntensityWavelen g th ScO/MgCaO/GaN Control Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 5-25 PL scans of samp les annealed non-radiated Sc2O3(20nm)/MgCaO(20nm)/GaN to 900C. A) Pl Spectra B) Bandedge inset.

PAGE 153

136 A300400500600700800 0.0 0.5 1.0 1.5 2.0 Peak IntensityWavelength ScO/MgCaO/GaN Control Rad 10MeV 1 year later Rad 10MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C B350360370380 0.0 0.5 1.0 Peak IntensityWavelength ScO/MgCaO/GaN Control Rad 10MeV 1 year later Rad 10MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 5-26 Sc2O3(20nm)/MgCaO(20nm)/GaN post radiation anneals to 900oC for radiation energies of 10MeV. A) PL spectra B) Bandedge inset.

PAGE 154

137 A300400500600700800 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 Peak IntensityWavelength ScO/MgCaO/GaN Control Rad 40MeV 1 year later Rad 40MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C B350360370380 0.0 0.5 1.0 Peak IntensityWavelength ScO/MgCaO/GaN Control Rad 40MeV 1 year later Rad 40MeV Annealed 200C Annealed 300C Annealed 400C Annealed 600C Annealed 800C Annealed 900C Figure 5Sc2O3(20nm)/MgCaO(20nm)/GaN post ra diation anneals to 900oC for radiation energies of 40MeV. A) PL spectra B) Bandedge inset.

PAGE 155

138 123 1 10 100 1000 10000 100000 1000000 Control Annealed 400CXRR Counts Figure 5-28 Sc2O3(20nm)/MgCaO(20nm)/GaN XRR scan of annealed sample.

PAGE 156

139 123 1 10 100 1000 10000 100000 1000000 Pre Rad Post 10MeV Rad Post 40MeV Rad Post 10MeV Rad Annealed to 400C Post 40Mev Rad Annealed to 400CXRR Counts Figure 5-29 Sc2O3(20nm)/MgCaO(20nm)/GaN XRR scans pre and post radiation annealed at 400oC.

PAGE 157

140 02004006008001000 0.0 0.2 0.4 0.6 0.8 1.0 Normalized Bandedge RatioTemperature Sc2O3/MgCaO/GaN Control Radiated 10MeV Radiated 40MeV Rad 10MeV 1Yr. Later Rad 40meV 1Yr. Later Figure 5-30 Normalized bandedge peaks of annealed Sc2O3(20nm)/MgCaO(20nm)/GaN to the Sc2O3(20nm)/MgCaO(20nm)/GaN control peak intensity of 1.2968 vs. annealing temperature.

PAGE 158

141 CHAPTER 6 PROCESSING AND THERMAL STABILITY OF MOS DEVICES The following sections will discuss the processing of novel gate dielectric semiconductor devices and the thermal stab ility of the fabricated MOS devices. 6.1 Processing of Novel Oxide/Gallium Nitride Devices Through much trial and error a process fo r fabricating single structure and stack structure MOS diodes was found. A mask set wa s designed to be used for a variety of MOS diodes. The mask set was developed usi ng Microsoft PowerPoint that was reduced to MicroFiche to acquire the desired devi ces sizes. Figure 6-1 shows the basic mask design, which had a standard ohmic pad a nd two different gate pad sizes of 50m and 100m. Two different methods of proces sing the MOS devices were studied. 1. A process in which ohmic contacts were deposited before oxide growth on GaN. 2. A process in which ohmic contacts were deposited after dielectric growth on GaN. The advantage of putting ohmic contacts on th e GaN before oxide growth, allows for annealing of the ohmic contacts without causi ng thermal damage to the oxide. However, once the ohmic contacts are on the GaN and anneal ed, the substrate should ideally go thru a dilute standard MBE substrate cleaning be fore growth of the oxide can be done. The problem that can arise with the standard cleaning, it is possible damage to the ohmic contacts via etching of the me tal from the buffered oxide etch used to clean the substrate surface. A simple ohmic contact study was pe rformed to verify which conditions would result in the best ohmic contacts. Four ohm ic samples with the following conditions were tested:

PAGE 159

142 1. RTA Anneal at 750oC for 30 seconds 2. No Pre Clean, in-situ MBE Anneal at 700oC for 5 minutes facing sources 3. Pre Clean with UV-Ozone and dilute BOE, in-situ MBE anneal at 700oC for 5 minutes facing sources. 4. UV-Ozone Treatment, in-situ MBE anneal at 700oC for 5 minutes facing sources. Figure 62 shows the IV results for each of the conditions. Based on the results the ohmic sample that received just the UV-Ozone treatment, showed a more linear profile after annealing and proved to be the better condition for ohmics before oxide growth. One problem that does exist with this approa ch is making contact with the electrical probes through the oxide to the ohmic pad. He nce it was necessary to find an etching recipe that would work for the stack structur es. Table 6-1 shows the recipes used to etch each oxide that was studied. The stack structure of Sc2O3/MgCaO or Sc2O3/MgO is dry etched through the scandium oxide and then wet etched through the MgO or MgCaO. For this study all the diode sample s were processed post oxide growth on u-GaN substrates. 6.2 Proton Radiation Effects of GaN MOS Diodes Radiation effects of different oxides and oxide structures were tested to determine which oxide would be the least susceptible to proton bombardment. Figures 6-3 and 6-4 show the IV and CV behavior of the Sc2O3/ GaN sample before and after radiation. Although there is a decrease th e turn-on has also become more abrupt. The CV shows approximately a positive 1V flatband voltage sh ift possibly indicating an increase in fixed oxide charge. In figures 6-5 and 6-6, the electrical behavior of the MgO stack structure oxide is presented. Both IV and CV show degr adation of the electri cal characteristics of the sample. In the current-voltage plot ther e is a decrease in br eakdown voltage and in capacitance-voltage, there is a small amount of change post i rradiation that is affecting

PAGE 160

143 the fixed oxide charge. For the MgCaO stack structure to samples with different oxides thicknesses where grown and tested to determ ine which thickness of MgCaO was better to have at the interface of the GaN device. As it is shown in figure 6-7, the sample with the 5nm MgCaO showed better breakdown than the 35nm MgCaO. From this point the 5nm MgCaO with capping layer was investigated for the radiation and thermal annealing. Figure 6-8 shows the IV breakdown of the samp le after 10MeV irradiation. As expected the breakdown decreases. This is keeping with the model that irradiation of GaN devices still experience some degradation after e xposure to protons. Figur e 6-9 shows the CV characteristics of the ternary oxide. As noted there is a large increase of forward flatband voltage shift, which has resulted in a poor CV curves. Inaccurate or poor data does not allow for calculations of interface state dens ity. Interface state densities(Dit) of the irradiated sample are as li sted in table 6-2. In figure 6-10, the Dit plots for the Sc2O3/GaN and the Sc2O3/MgO/GaN samples. The degradation of the ternary oxide was too poor to resolve a dit measurement from. 6.3 Effects of In-situ and Ex-situ Thermal Annealing Scandium oxide was tested to determine th e electrical characteristics for use as a capping layer for MgO and MgCaO. Figure 6-11 shows the IV results of the Sc2O3/GaN during heating on the hot chuck of the probe st ation. The measurements were taken as the sample was heated from 100oC to 300oC, typical operating temper atures for GaN devices. Notice that while the sample is hot, the breakdo wn is less than when the sample as been cooled back to room temperature. The figur e also shows an incr ease in breakdown with after the final cooling back to room temperature after being heated to 300oC. After 100oC no relevant CV data could be co llected as shown in figure 6-12.

PAGE 161

144 A significant problem with MgO that has b een observed in the past, is the slow degradation of the oxide in ai r. Over a period of time the oxide begins to deteriorate through the formation of hydroxyl speci es. Hence the reason for using Sc2O3 as a capping layer to the MgO, to prevent the degradation of the sample while it is exposed to air, resulting in 8nm of Sc2O3 capping layer and 40nm of MgO on u-GaN. This sample was grown at a low oxygen pressure of .2 torr-1, which has shown to have better electrical characteristics than oxides grown at highe r oxygen pressures. Figure 6-13 shows the thermal behavior for IV testing of the samp le. Thermal cycling of the sample decreases its breakdown field and during heating, there was no relevant CV during the heating of the sample. This implies that there are longtime traps that are not reaching equilibrium while the sample is heated. Figure 6-14 s hows the post heating CV behavior. Post 100oC there is some slight improvement of the flat band voltage shifts but this worsens as the sample is heated and cooled again over 200oC to 300oC, as seen by the forward shift in flat band voltage. Magnesium calcium oxide (MgCaO) was grown to determine the electric characteristics since MgCaO has a lattice match th at is very close to the lattice of GaN. The idea behind this is that the better the lattice matching the smaller interface defect density. High interface defect densities create current leakages, which affect the threshold gate voltage, Vth. All devices using MgCaO are capped with scandium oxide to prevent degradation of the MgCaO. The most critical layers of the diel ectric occur several monolayers at the interface. 5nm of MgCaO followed by 35nm Sc2O3 was grown on one sample and 35nm of MgCaO with 5nm Sc2O3 cap was grown on the second sample. Both of the samples were fabricated into MOS diodes and tested. As shown previously in

PAGE 162

145 figure 6-7 the current-voltage measurements of the two different di electric thicknesses of the MgCaO. From the data shown the Mg CaO breakdown voltage for both samples is roughly .75 to 1 V. The low breakdown suggest s that even thought the oxide is maor lattice matched there is stil l a lot of current leakage through the oxide. Two possible reasons are: 1. Defects in the oxide: change s in the epitaxial growth process may improve the interface, which could in turn improve the breakdown voltage. 2. Residual water: since it is known that Mg CaO readily hydrolyzes in water, care should be taking during the fabrication pr ocess to dehydrate the sample via low temperature furnace bakes after water is used on the sample. These few changes could very well make a significant difference in the electrical measurements of the MgCaO devices. The sample with the 5nm MgCaO oxide was further tested during and after heating on the hot chuck probe station. Figures 6-15 through 6-17 show the results of th e IV and CV prof iles of the 35nm Sc2O3/5nm MgCaO/u-GaN sample. As shown in the IV grap h, after the initial de crease in breakdown voltage the sample under goes very small changes upon continued thermal cycling. The CV shows shifts in the flat band voltage indicating a cha nge in fixed oxide charge. Figure 6-18 shows a separation of heated CV and cooled room temperature CV for better comparison purposes. Forming gas (H2+N2) anneals were also performed on each sample in an RTA at 400oC for 30 seconds. The purpose was to determine if the anneal would improve the samples characteristics. Figures 6-18 to 620 present the CV behavior post forming gas anneal. Of the 3 samples the the 35nm Sc2O3/5nm MgCaO/u-GaN showed a decrease in fixed oxide charge based on the negative flat ba nd voltage shift from the initial CV curve. But the calculated Dit for the forming gas a nneals show just the opposite for the samples.

PAGE 163

146 Dit indicates that the Sc2O3/MgO/GaN is the better samp le but judging by the shape of the CV curve this shouldn’t hold to be true. This type of erro r can be attribut ed to errors in measurements or calculation of the interf ace state density. The Dit of each annealed sample is shown in figures 6-21 to 6-22. 6.4 Summary In short radiation and ann ealing of the MOS diodes does degrade the samples, just as it does with the PL in the gallium nitride materials. It also causes a forward flatband voltage shift in the CV curves, indicating high er fixed oxide charge and a decrease in IV breakdown.

PAGE 164

147 Table 6-1 Oxide Etchants and Condition. Oxide Wet Etch Solution Dry Etching Recipe Time Sc2O3 ICP CL Etching 20mins MgO 2% Phosphoric Acid/DI Water 15 to 17 secs. MgCaO 2% Phosphoric Acid/DI Water 15 secs. Sc2O3/MgCaO or MgO 2% Phosphoric Acid/DI Water ICP CL Etching through Sc2O3

PAGE 165

148 Table 6-2 Dit Values at .4eV for Oxide/GaN Devices 40nmSc2O3/GaN Dit @ .4eV Control 5.955E11 10MeV Radiation 1.03E12 Forming Gas Anneal @ 400C 7.82E11 In-situ 100C 1.065E12 Post 100C 6.949E11 2nm Sc2O3/40nm MgO/GaN Dit @ .4eV Control 2.33E11 10MeV Radiation 5.288E11 Forming Gas Anneal @ 400C 1.03E11 Post 100C 2.79E11 Post 200C 2.65E11 Post 300C 2.39E11 35nm Sc2O3/4nm MgCaO/GaN Dit @ .4eV Control 2.21E11 Forming Gas Anneal @ 400C 1.17E12 In-situ 100C 6.39E11 Post 100C 1.99E11 In-situ 200C 5.30E11 Post 200C 8.65E11 In-situ 300C 6.289E11 Post 300C 1.29E12

PAGE 166

149 Figure 6-1 Finished diode mask design.

PAGE 167

150 0.40.60.81.01.21.41.61.82.02.22.4 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10 0.11 RTA Anneal NO Pre Treatment, MBE Anneal Pre Treatment, MBE Anneal UV-Ozone, MBE AnnealCurrentVoltage Figure 6-2 Ohmic IV plots taken from samples receiving various treatments.

PAGE 168

151 -6-4-20246810 0.000 0.001 0.002 0.003 0.004 0.005 0.006 Current DensityVoltage Sc2O3 Control 2 Rad 10MeV Figure 6-3 Current density –Voltage behavior of Sc2O3/GaN after 10MeV irradiation.

PAGE 169

152 -4-3-2-101234 0.00E+000 2.00E-013 4.00E-013 6.00E-013 8.00E-013 1.00E-012 1.20E-012 1.40E-012 1.60E-012 1.80E-012 2.00E-012 2.20E-012 2.40E-012 CapacitanceVoltage Sc2O3/GaN Control Rad 10MeV Figure 6-4 Capacitance –Voltage plot of Sc2O3/GaN after 10MeV irradiation.

PAGE 170

153 -4-20246810 -0.0002 0.0000 0.0002 0.0004 0.0006 0.0008 0.0010 0.0012 0.0014 Current DensityVoltage Sc2O3/MgO/GaN Control Rad 10MeV Figure 6-5 Current density –Voltage behavior of Sc2O3/ MgO/GaN after 10MeV irradiation.

PAGE 171

154 -4-202468 0.00E+000 2.00E-012 4.00E-012 6.00E-012 8.00E-012 1.00E-011 CapacitanceVoltage Sc2O3/MgO/GaN Control Rad 10MeV Figure 6-6 Capacitance –Voltage plot of Sc2O3/ MgO/GaN after 10MeV irradiation.

PAGE 172

155 -3-2-1012 0.00E+000 2.00E-008 4.00E-008 6.00E-008 8.00E-008 1.00E-007 CurrentVoltage MgCaO 5nm MgCaO 35nm Figure 6-7 Comparison of Current Dens ity-Voltage results of different Sc2O3/MgCaO/GaN thickness.

PAGE 173

156 -6-5-4-3-2-10123 -0.0002 0.0000 0.0002 0.0004 0.0006 0.0008 0.0010 0.0012 0.0014 Current DensityVoltage Sc2O3/5nm MgCaO/GaN Control Rad 10MeV Figure 6-8 Capacitance-Voltage plot of Sc2O3/MgCaO/GaN after 10MeV irradiation.

PAGE 174

157 -3.5-3.0-2.5-2.0-1.5-1.0-0.50.00.51.01.52.02.53.03.54.04.5 0.00E+000 2.00E-012 4.00E-012 6.00E-012 8.00E-012 1.00E-011 1.20E-011 1.40E-011 CapacitanceVoltage Sc2O3/5nm MgCaO/GaN Control Rad 10MeV Figure 6-9 Capacitance-Voltage plot of Sc2O3/MgCaO/GaN after 10MeV irradiation.

PAGE 175

158 A0.10.20.30.40.50.60.7 5.00E+011 1.00E+012 1.50E+012 2.00E+012 2.50E+012 3.00E+012 3.50E+012 DitEc-Et 40nm Sc2O3/GaN Control 10MeV Rad B0.150.200.250.300.350.400.450.500.550.600.65 2.00E+011 4.00E+011 6.00E+011 8.00E+011 1.00E+012 DitEc-Et 2nm Sc2O3/40nm MgO/GaN Control 10MeV Rad Figure 6-10 Dit of post 10MeV radiation. A) Sc2O3/GaN B) Sc2O3/MgO/GaN

PAGE 176

159 -12-10-8-6-4-2024681012 0.000 0.001 0.002 0.003 0.004 0.005 0.006 Current DensityVoltage Sc2O3 Control In Situ Anneal 100C Cooled from 100C In Situ Anneal 200C Cooled from 200C In Situ Anneal 300C Cooled from 300C Figure 6-11 Current Density –Voltage behavior of Sc2O3/GaN during and after thermal annealing.

PAGE 177

160 -4-3-2-1012345 0.00E+000 2.00E-013 4.00E-013 6.00E-013 8.00E-013 1.00E-012 1.20E-012 1.40E-012 1.60E-012 1.80E-012 2.00E-012 2.20E-012 2.40E-012 2.60E-012 2.80E-012 3.00E-012 3.20E-012 CapcitanceVoltage Sc2O3 Control In Situ 100C Cooled from 100C Figure 6-12 Capacitance –Voltage plot of Sc2O3/GaN during and after heating.

PAGE 178

161 -6-4-20246810 -0.0002 0.0000 0.0002 0.0004 0.0006 0.0008 0.0010 0.0012 0.0014 Current DensityVoltage Sc2O3/MgO/GaN Control In-situ 100C Cooled from 100C In-situ 200C Cooled from 200C In-Situ 300C Cooled from 300C Figure 6-13 Current Density –Voltage behavior of Sc2O3/MgO/GaN during and after heating.

PAGE 179

162 -4-202468 0.00E+000 2.00E-013 4.00E-013 6.00E-013 8.00E-013 1.00E-012 1.20E-012 1.40E-012 1.60E-012 1.80E-012 2.00E-012 2.20E-012 2.40E-012 2.60E-012 CapacitanceVoltage Sc2O3/MgO/GaN Control Post 100C Post 200C Post 300C Figure 6-14 Capacitance –Voltage breakdown of Sc2O3/MgO/GaN after heating.

PAGE 180

163 -6-5-4-3-2-10123 -0.0002 0.0000 0.0002 0.0004 0.0006 0.0008 0.0010 0.0012 0.0014 Current DensityVoltage Sc2O3/50A MgCaO/GaN Control In-situ 100C Cooled from 100C In-situ 200C Cooled from 200C In-Situ 300C Cooled from 300C Figure 6-15 Current Density –Voltage behavior of Sc2O3/5nm MgCaO/GaN during and after heating.

PAGE 181

164 -4.5-4.0-3.5-3.0-2.5-2.0-1.5-1.0-0.50.00.51.01.52.02.53.03.5 0.00E+000 2.00E-012 4.00E-012 6.00E-012 8.00E-012 1.00E-011 1.20E-011 1.40E-011 1.60E-011 1.80E-011 2.00E-011 2.20E-011 2.40E-011 2.60E-011 CapacitanceVoltage Sc2O3/50A MgCaO/GaN In-situ 100C Cooled from 100C In-situ 200C Cooled from 200C In-situ 300C Cooled from 300C Figure 6-16 CapacitanceVoltage plot of Sc2O3/MgCaO/GaN during and after heating.

PAGE 182

165 A-4.5-4.0-3.5-3.0-2.5-2.0-1.5-1.0-0.50.00.51.01.52.02.53.03.5 0.00E+000 2.00E-012 4.00E-012 6.00E-012 8.00E-012 1.00E-011 1.20E-011 1.40E-011 1.60E-011 1.80E-011 2.00E-011 2.20E-011 2.40E-011 2.60E-011 CapacitanceVoltage Sc2O3/50A MgCaO/GaN In-situ 100C Cooled from 100C In-situ 200C Cooled from 200C In-situ 300C Cooled from 300C B-4.5-4.0-3.5-3.0-2.5-2.0-1.5-1.0-0.50.00.51.01.52.02.53.03.5 0.00E+000 2.00E-012 4.00E-012 6.00E-012 8.00E-012 1.00E-011 1.20E-011 1.40E-011 1.60E-011 1.80E-011 2.00E-011 2.20E-011 2.40E-011 2.60E-011 CapacitanceVoltage Sc2O3/50A MgCaO/GaN In-situ 100C Cooled from 100C In-situ 200C Cooled from 200C In-situ 300C Cooled from 300C Figure 6-17 CapacitanceVoltage plot of Sc2O3/MgCaO/GaN during and after heating. A) CV of in-situ testing. B) CV of sa mples cooled to room temperature.

PAGE 183

166 -3-2-10123 0.00E+000 2.00E-013 4.00E-013 6.00E-013 8.00E-013 1.00E-012 1.20E-012 1.40E-012 1.60E-012 1.80E-012 CapacitanceVolta g e Sc2O3/GaN Control Forming Gas Anneal at 400C Figure 6-18 Capacitance –Voltage plot of Sc2O3/GaN after forming gas anneal at 400oC.

PAGE 184

167 -4-202468 0.00E+000 2.00E-013 4.00E-013 6.00E-013 8.00E-013 1.00E-012 1.20E-012 1.40E-012 1.60E-012 1.80E-012 CapacitanceVoltage Sc2O3/MgO/GaN Control Forming Gas Anneal at 400C Figure 6-19 Capacitance –Voltage plot of Sc2O3/MgO/GaN after forming gas anneal at 400oC.

PAGE 185

168 -2.5-2.0-1.5-1.0-0.50.00.51.01.52.0 0.00E+000 2.00E-012 4.00E-012 6.00E-012 8.00E-012 1.00E-011 1.20E-011 1.40E-011 1.60E-011 1.80E-011 2.00E-011 2.20E-011 CapacitanceVoltage Sc2O3/50A MgCaO/GaN Control Forming Gas Anneal at 400C Figure 6-20 CapacitanceVoltage plot of Sc2O3/MgCaO/GaN after forming gas anneal at 400oC.

PAGE 186

169 A0.10.20.30.40.50.60.7 5.00E+011 1.00E+012 1.50E+012 2.00E+012 2.50E+012 DitEc-Et 40nm Sc2O3/GaN Control In-situ 100C Cooled from 100C B0.10.20.30.40.50.60.7 5.00E+011 1.00E+012 1.50E+012 2.00E+012 2.50E+012 DitEc-Et 40nm Sc2O3/GaN Control Forming Gas Anneal @ 400C Figure 6-21 Dit of Annealed Sc2O3/GaN. A) Probe station thermal anneals. B) Forming Gas anneal.

PAGE 187

170 A0.10.20.30.40.50.60.70.8 2.00E+011 4.00E+011 6.00E+011 8.00E+011 1.00E+012 1.20E+012 1.40E+012 DitEc-Et 2nmSc2O3/40nmMgO/GaN Control Post 100C Post 200C Post 300C B0.00.20.40.60.8 0.00E+000 5.00E+011 1.00E+012 1.50E+012 2.00E+012 2.50E+012 DitEc-Et 2nmSc2O3/40nmMgO/GaN Control Forming Gas Anneal @ 400C Figure 6-22 Dit of Annealed 2nm Sc2O3/40nm MgO/GaN. A) Probe station thermal anneals. B) Forming gas anneal.

PAGE 188

171 A0.100.150.200.250.300.350.400.450.500.550.600.650.700.750.800.85 0.00E+000 5.00E+011 1.00E+012 1.50E+012 2.00E+012 2.50E+012 DitEc-Et Sc2O3/MgCaO Control In-Situ 100C Post 100C In-Situ 200C Post 200C In-Situ 300C Post 300C B0.00.20.40.60.8 0.00E+000 5.00E+011 1.00E+012 1.50E+012 2.00E+012 2.50E+012 DitEc-Et 2nmSc2O3/40nmMgO/GaN Control Forming Gas Anneal @ 400C Figure 6-23 Dit of Annealed 35nm Sc2O3/5 nm MgCaO/GaN A) Probe station thermal anneals. B) Forming gas anneal.

PAGE 189

172 CHAPTER 7 SUMMARY AND DISSCUSSION 7.1 GaN vs. Oxide/GaN Based on the data collected, some general trends for the GaN can be noted. First is that radiation and annealing of the samples with and without oxides, decrease the PL luminescence intensity, except the activated P-GaN. Radiation of the activated P-GaN sample enhances the PL emission. Second, an increase of PL intensity at 200oC has shown to be unique to the UOE GaNs and activated P-GaN. After approximately 400oC for non-oxide samples, the drastic decline in peak intensity levels off and become less significant in change with hi gher temperatures. Where as the oxide/GaN samples shows an extended range of PL intensity increase th an that of the non-oxide GaN samples. The increase in PL intensity after irra diation and annealing of the n-GaN and activated P-GaN can possibly be explained by trap filling and emptying. In n-type GaN before radiation or annealing the traps are filled but do not contribute to PL emission. Once the sample is radiated the trap becomes empty beco ming a non-radiative center and takes a long time to relaxed back to its filled neutral state. This long time relaxation is shown in figure 7-1, which presents a PL scan of Sc2O3/MgCaO/GaN seven months after 900oC anneal. The plot shows that over a peri od of time the luminescence of the GaN is recoverable to a certain poi nt. Although it shows recover y, it does not recover to the initial PL intensity of 1.29. For the activated P-GaN, the traps ar e initially empty and non-radiative due to the location of the fermi level within the sample. After irradiation these traps are filled with a carriers and the PL increases. Once the trap has had ample

PAGE 190

173 time to relax it should return to its empty non -radiative state thus re sulting in lower PL intensity. The empty traps compete for carrier s therefore reducing the amount of carriers available for radiative transitions. Based on behavi or of p-GaN, trap level is estimated to be at ~1.655eV above valence band which is in agreement with literature reports using low energy proton radiation. The roughening of the oxide/GaN interface s uggests that operating temperatures of any device fabricated from the MgO and MgCaO oxides maybe be limited. As with the non-oxide GaN samples the irradiation of the oxide samples still significantly decreases the bandedge emission and increases the de fect emission. The radiation also causes roughening of the oxide/semiconductor interface and roughening at the surface. After one year, the samples show some relaxation of long life-time traps that degrade the PL emission, as was shown with the non-oxide Ga N. The effect of annealing on the oxide also causes samples to roughening at the surf ace and interface. Radiation of the samples makes them more susceptible to annealing damage because of the already present defects. 7.2 MOS Devices Radiation and annealing of the MOS diodes does degrade the samples, just as it does with the PL and XRR in the gallium nitr ide materials. It also causes a forward flatband voltage shift in the CV curves, indicating higher fixed oxide charge and a decrease in IV breakdown. Radiation and ann ealing empties traps in both the bulk GaN and the oxide, resulting in higher Dit and an in crease in diode leakag e. An increase in diode current leakage causes a decrease in forward IV breakdown. With the CV curves the forward flatband voltage shif ts indicate an increase in fi xed oxide charge and damage to the dielectric. Based on the study, the oxide s do not protect the GaN from bulk damage from the radiation or annealing. What has been observed with the oxi des is that the Mg-

PAGE 191

174 containing oxides showed better resistance than Sc2O3 to radiation and thermal annealing. As stated in earlier sections, the purpose of the alternate diel ectrics was to investigate one that allowed for elevated operating temper atures and stability harsh environments.

PAGE 192

175 300400500600700800 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 Peak IntensityWavelength Sc2O3/MgCaO Annealed 900C Post 900C 7 months later Figure 7-1 Recovery of Oxide/GaN post 900oC after 7 months.

PAGE 193

176 LIST OF REFERENCES [1] B.G. Streeman, Solid State Electronic Devices Englewood Cliffs, NJ: Prentice Hall, 1995. [2] S. M Sze, Semiconductor Devices. Murray Hill NY: John Wiley and Sons, 1985. [3] R.E. Hummel, Understanding Materials Science. New York, NY: Springer, 1999. [4] D.R. Askeland, The Science and Engineering of Materials. 3rd ed. Boston, MA: PWS 1994. [5] R. Wolfson, J.M. Pasachoff, Physics. Boston, MA: Little, Brown and Company; 1987. [6] H.C. Casey Jr, G.G. Fountain, R.G. Alley, B.P Keller, S.P. Denbaars, Applied Physics Letters vol. 68, 1850, 1996. [7] M. Sawada, T. Sawada, Y. Yamagata, K. Imai, H. Kumura, M. Yoshino, K. Iizuka, H. Tomozawa, Proceedings of the Second International Conference on Nitride Semiconductors Tokushima, 482, 1997. [8] S. Aurlkumaran, T. Egawa, H. Ishikawa, T. Jimbo, M. Umeno, Applied Physics Letters vol. 73, 809-811,1998. [9] S.L. Rumyanysev, N. Pala, M.S. Shur, R. Gaska, M.E. Levinshtein, M. Asif Khan, G. Simin, X. Hu, J. Yang, Journal of Applied Physics vol. 90, 310-314, 2001. [10] T.E. Cook, Jr. C.C. Fulton, W.J. Mecouc h, K.M. Tracey, R.F. Davis, E.H. Hurt, G. Lucovsky, R.J. Nemanich, Journal of Applied Physics vol 93, 3995-4004, 2003. [11] K.M. Chang, C.C. Cheng, C.C. Lang, Solid State Electronics vol.46 1399-1403, 2002. [12] T. Hashizumu, S. Ootomo, T. Inagaki, H. Hasegawa, Journal of Vacuum Science and Technology B vol. 21, 1828-1830, 2003. [13] T.P. Ma, Microelectronics Journal vol. 34 363-370, 2003. [14] B. Gaffey, G. Chong, L. Guido, X.W. Wang, T.P. Ma, Applied Physics Letters vol. 42, 2145, 1998.

PAGE 194

177 [15] F. Ren, C.R. Abernathy, J.D. MacKenzi e, B.P. Gila, S.J. Pearton, M. Hong, M.A. Marcus, M.J. Schuman, A.G. Baca, R.J. Shul, Solid State Electronics vol. 42, 2177, 1998. [16] H. Kawai, M. Hara, F. Nakamura, T. Asatsuma, T. Kobayashi, S. Imanaga, Journal of Crystal Growth vol.189/190, 738, 1998. [17] S.D. Wolter, B.P. Luther, D.L. Waltermyer, C. Onneby, S.E. Mohney, R.J. Molnar, Applied Physics Letters vol.70, 2156, 1997. [18] S.D. Wolter, S.E. Mohney, H. Venug opalan, A.E. Wickenden, D.D. Koleske, Journal of the Elect rochemical Society vol.145, 629, 1998. [19] Y. Nakano, T. Kachi, T. Jimbo, Journal of Vacuum Science and Technology B vol. 21, 2220-2222, 2003. [20] M. Hong, J. Kwo, S.N.G. Chu, J.P. Ma nnaerts, A.R. Kortan, H.M. Ng, A.Y. Cho, K.A. Anselm, C.M. Lee, J.I. Chyi, Journal of Vacuum Science and Technology B vol. 20, 1274-1277, May/June 2002. [21] D.J. Fu, Y.H. Kwon, C.J. Park, K.H. B aek, H.Y. Cho, D.H. Shin, C.H. Lee, K.S. Chung, Applied Physics Letters vol.80, 446-448, 2002. [22] L.H. Peng, C.H. Liao, Y.C. Hsu, C.S. Jong, C.N. Huang, J.K. Ho, C.C. Chiu, C.Y. Chen, Applied Physics Letters vol. 76, 511, 2000. [23] D.J. Fu, T.W. Kang, Sh.U. Yuldashev, N. H. Kim, S.H. Park, J.S. Yun, K.S. Chung, Applied Physics Letters vol.78, 1309, 2001. [24] C. T. Lee, H.W. Chen, H.Y. Lee, “Metal-oxide-semiconductor devices using Ga2O3 dielectrics on n-type GaN”, Applied Physics Letters vol. 82(24), 4304-4306, 2003. [25] T. Rotter, R. Ferretti, D. Mistele, F. Fedler, H. Klausing, J. Stemmer, O.K. Semschinova, J. Aderhold, J. Graul, Journal of Crystal Growth vol. 230, 202-206, 2001. [26] C. Bae, G.B. Rayner, G. Lucovsky, Applied Surface Science, vol. 216, 119-123, 2003. [27] R. Therriem. G. Lucovsky, R. Davis, Applied Surface Science, vol.166, 513-519, 2000. [28] C. Bae, G. Lucovsky, Surface Science, vol. 532-535, 759-763, 2003. [29] C. Bae, G. Lucovsky, Applied Surface Science vol. 212-213, 644-648, 2003. [30]Y. Nakano, T. Kachi, T. Jimbo, Applied Physics Letter vol. 83, 4336-4338, 2003.

PAGE 195

178 [31] S.J. Chang, Y.K. Su, Y.Z. Chiou, J.R. Ciuo, B.R. Huang, C.S. Chang, J.F. Chen, Journal of the Elect rochemical Society vol.150, C77-C80, 2003. [32] E.D. Readinger, S.D. Wlter, D.L. Waltermyer, J.M. Delucca, S.E. Mohney, B.I. Prenitzer, L.A. Gainnuzzi, R.J. Molnar, Journal of Electronic Materials vol. 28, 257,1999. [33] M. Hong, J.P. Mannaerts, J.E. Bower, J. Kwo, M. Passlack, W-Y. Hwang, L.W. Tu, Journal of Crystal Growth vol.175/176, 422, 1997. [34] F. Ren, M. Hong, W.S Hobson, J.M. K uo, J.R. Lothian, J.P. Mannaerts, J. Kwo, S.N.G. Chu, Y.K. Chen, A.Y. Cho, Solid State Electronics ,vol. 41, 1751, 1997. [35] Y.C. Wang, M. Hong, J.M. Kuo, J.P Mannaer ts, J. Kwo, H.S. Tsai, J.J. Krajewski, J.S. Weiner, Y.K. Chen, A.Y. Cho, Materials Research Society Symposium Proceedings vol. 573, 219, 1999. [36] J. Kwo, M. Hong, A.R Kortan, D.W. Mu rphy, J.P Mannaerts, A.M. Sergent, Y.C. Wang, K.C Hsieh, Materials Research Society Symposium Proceedings, vol. 573, 57, 1999. [37] T.S. Lay, M. Hong, J. Kwo, J.P Mannaerts, W.H. Hung, D.J. Huang, Materials Research Society Symposium Proceedings, vol. 573, 131, 1999. [38] A.R. Mortan, M. Hong, J. Kwo, J.P Mannaerts, N. Kopylov, Materials Research Society Symposium Proceedings, vol. 573, 21, 1999. [39] F. Ren, M. Hong, S.N.G. Chu,, M.A. Marcus, A. Baca, S.J. Pearton, C.R. Abernathy, Applied Physics Letters vol. 73, 3893-3895, 1998. [40] J. Kim, R. Mehandru, B. Lou, F. Ren, B.P. Gila, A.H. Onstine, C.R. Abernathy, S.J. Pearton, Y. Irokawa, Applied Physics Letters 81 (2), 373-375, 2002. [41] B. Lou, J.W Johnson, B.P. Gila, A.H. Onstine, C.R. Abernathy, F. Ren, S.J. Pearton, A.G. Baca, A.M. Dabiran, A.M. Wowchack, P.P. Chow, Solid State Electronics vol.46, 467-476, 2002. [42] R. Mehandru, B.P. Gila, J. Kim, J.W Johnson, K.P. Lee, B. Lou, A.H. Onstine, C.R. Abernathy, S.J. Pearton, F. Ren, Electrochemical and Solid State Letters vol. 5, G51-G53, 2002. [43] B. Lou, J.W Johnson, J. Kim, R. Mehandr u, F. Ren, B.P. Gila, A.H. Onstine, C.R. Abernathy, S.J. Pearton, A.G. Baca, R.D. Briggs, R.J. Shul, C. Monier, J. Han, Applied Physics Letters vol. 80, 1661-1663, 2002. [44] J. Kim, B.P. Gila, R. Mehandru, J.W Johnson, J.H. Shin, K.P. Lee, B. Lou, A.H. Onstine, C.R. Abernathy, S.J. Pearton, F. Ren, Journal of the Electrochemical Society vol. 149, G482-G484, 2002.

PAGE 196

179 [45] S.A. Campbell, The Science and Engeering of Microelectronic Fabrication. New York, NY: Oxford University Press, 1996. [46] W.N. Hess, The Radiation Belt and Magnetosphere Blaisdell Publ. Co.,NY 1968. [47] E.J. Daly, ESA Journal vol.12, 229, 1988. [48] T.P. Ma, P.V. Dressendorfer Ionizing Radiation Effe cts in MOS Devices and Circuits NY: John Wiley and Sons, 1989. [49] B. Luo, J.W. Johnson, F. Ren, K. Allu ms, CR Abernathy, SJ Pearton, AM Dabiran, AM Wowchack, CJ Polley, PP Chow, D Schoenfield, AG Baca, Applied Physics Letters. 80(4), 604-6, 2002. [50] D.C. Look, D.C. Reynolds, J.W. Hemsky, J. R. Sizelove, R.L. Jones, R.J. Molnar, Physics Review Letters. vol. 79 num.12, 1997. [51] M .Osinki, P. Perlin, H. Schone, A.H. Paxton, E.W. Taylor, Electronics Letters vol. 33, num. 14, 1997. [52] S.M. Khanna, J. Webb, H. Tang, A.J. Houdayer, C. Carlone, IEEE Trans. Nucl. Sci. vol. 47 num 6, 2322, 2000. [53] S.J. Cai, Y.S. Tang, R Li, YY Wei, L Wong, YL Chen, KL Wang, M Chen, YF Zhao, YD Schrimpf, JC Keay, K.F. Galloway, IEEE Trans. Electron Devices vol. 47 num.2, 304, 2000. [54] VV Emtsev, VYu Davydov, EE Halle r, AA Klochikhin, VV Kozlovskii, GA Oganesyan, DS Poloskin, NM Shmidt, VA Vekshin, AS Usikov, Physica B 308310 2001. [55] E Gaubas, S Jursenas, R Tomasiunas, J Vaitkus, A Zukauskas, A Blue, M Rahman, KM. Smith, Nuclear Instruments and Met hods in Physics Research A vol. 546, 247-251 2005. [56] R Mehandru, BP Gila, J Kim, JW Johnson, KP Lee, B Luo, AH Onstine, CR Abernathy, SJ Pearton, F Ren, MRS Conference Proceedings ,1040 Boston, MA, Nov. 26-30 2001. [57] C. T. Sah, Fundamentals of Solid -State Electronics ,World Scientific, New Jersey, 1991. [58] C.R. Brundle, C.A. Evans Jr., S.Wilson, editors, Encyclopedia of Materials Characterization Greenwich, CT: Manning, 1992. [59] A. Gibaud, S. Hazra. Current Science. vol. 78, no. 12, 1467-1477, June 25 2000.

PAGE 197

180 [60] KK Allums, B Luo, R Mehandru, BP Gila R Dwivedi, TN Fogarty, R Wilkins, CR Abernathy, F Ren, SJ Pearton. MRS Conference Proceedings ,713 Boston, MA, Nov. 26-30 2001.

PAGE 198

181 BIOGRAPHICAL SKETCH Kimberly Karlotta Allums was born Ap ril 27, 1978, in Houston, Texas. After completing high school in May of 1996, she went on to attend Prairie View A&M University. Just four years later, on May 13, 2000, she received her Bachelor of Science degree in electrical engineering. She decided to continue he r education at the Universi ty of Florida after being awarded a GEM Fellowship and an NSF Fellowship (AGEP). In the fall of 2000 Kimberly began pursuing a Master of Sc ience degree in materials science and engineering with a specializa tion in electronic materials. After receiving her master’s degree in May 2002, she continued her graduate studies to complete a PhD in the same major. Kimberly was involved in sports and various organizations throughout her high school and collegiate years. She played volle yball for Prairie View A&M University for four years. She has also worked for the Univ ersity of Florida Intramural Sports program as a basketball official and sports supervisor for the last 3 years. She is currently a member of Zeta Phi Beta Sorority Inc., Ep silon Gamma Iota, Inc. (a co-ed engineering fraternity), and University of Florid a’s Black Graduate Student Organization. Upon graduation, her future goals include wo rking in the research and development industry, a professorship and eventu ally owning her own businesses.


Permanent Link: http://ufdc.ufl.edu/UFE0013123/00001

Material Information

Title: Proton Radiation and Thermal Stability of Gallium Nitride and Gallium Nitride Devices
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: UFE0013123:00001

Permanent Link: http://ufdc.ufl.edu/UFE0013123/00001

Material Information

Title: Proton Radiation and Thermal Stability of Gallium Nitride and Gallium Nitride Devices
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: UFE0013123:00001


This item has the following downloads:


Full Text












PROTON RADIATION AND THERMAL STABILITY
OF GALLIUM NITRIDE AND GALLIUM NITRIDE DEVICES















By

KIMBERLY K. ALLUMS


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

Kimberly K. Allums

































This document is dedicated to my Family and Friends. Thank you for your continued
support of my endeavors and most of all me.















ACKNOWLEDGMENTS

First and foremost I'd like to thank God, because without Him I wouldn't be where

I am today. I'd like to thank my family for being my support throughout my entire life

and always pushing me to do my best.

I'd also like to thank Dr. Cammy R. Abernathy for accepting me into her group and

for her guidance through my graduate curriculum. I thank my committee members for

their guidance and suggestions.

I especially thank Dr. Brent Gila for all his tutelage during my research work. I

thank Abernathy Electronics Group for their friendship and help with my research. I'd

like to thank Dr. Richard Wilkins, Mr. Shoja Ardalon and Mr. Dwevdi of Prairie View

A&M University for collaborating with us on this project. I appreciate all that every one

has done.

This work was supported by ONR Grant No. US Navy N00014-98-1-0204, and the

Prairie View A&M University work was supported by NASA Grant Nos. NCC 9-114.
















TABLE OF CONTENTS



A C K N O W L E D G M E N T S ................................................................................................. iv

L IS T O F T A B L E S .................................................................... ......... .... ....... ....... v iii

LIST OF FIGURES ......... ......................... ...... ........ ............ ix

ABSTRACT .............. .......................................... xvi

CHAPTER

1 IN T R O D U C T IO N ............................................................................. .............. ...

2 BACKGROUND AND LITERATURE REVIEW ..................... ........................... 3

2 .1 In tro d u ctio n ............................................................................... 3
2.2 Device Structures ................. ......... ........ .... .......... .. .. .... ..........3
2.2.1 Metal Oxide Semiconductor Field Effect Transistor ...............................3
2.2.2 Metal Oxide Semiconductor Capacitors......................... ..................5
2.3 Growth and Photoluminescence Spectroscopy of Gallium Nitride........................6
2.4 D ielectrics ..................................... ................................ ......... 10
2 .4 .1 Ideal D ielectrics......... ................................................ ........ ... ......... 10
2.4.2 Crystalline vs. Amorphous ...................................................................... 11
2.4.3 Present State of Dielectrics for Gallium Nitride............... ... ...............12
2.4.3.1 Silicon oxide on G aN ..................................................................... 13
2.4.3.2 Silicon nitride on G aN ..................................... ...... ............... 14
2.4.3.3 A lum inum nitride on G aN ............................................ ..............14
2.4.3.4 Gallium oxide on GaN .................................... .. ............... 15
2.4.3.5 Silicon dioxide on gallium oxide on GaN......................................16
2.4.3.6 Gallium gadolinium oxide on GaN ..............................................17
2.4.3.7 G adolinium oxide on G aN ........................................... ............17
2.4.3.8 Scandium oxide on GaN ...................................... ............... 17
2.5 Epitaxal G row th............................................................ ... ... .... 18
2.5.1 M olecular B eam Epitaxy ...................................... ................. .... .......... 19
2.5.2 Substrate Preparation ............................................ ........... ............... 19
2.6 Types and Effects of Radiation on Microelectronics .........................................21
2.7 Previous R radiation Studies ......................................................... .. ..........23




v









3 EXPERIM ENTAL APPROACH ........................................ .......................... 35

3 .1 A ltern ativ e D ielectrics .................................................................................3 5
3.2 O xide G row th Param eters.......................................................... ............... 36
3.2.1 Scandium Oxide G row th ........................................ ........................ 36
3.2.2 M agnesium Oxide Grow th ........................................ ...... ............... 36
3.2.3 Magnesium Calcium Oxide Growth................................ ...............36
3.3 Materials Characterization................... ......... ............................ 37
3.3.1 Auger Electron Spectroscopy ............. ............................... ...............37
3.3.2 A tom ic Force M icroscopy .................................... ................................... 37
3.3.3 Transmission Electron Microscopy..........................................................38
3.3.4 Scanning Electron M icroscopy........................................ .....................38
3.3.5 X -R ay R eflectivity........................................................... ............... 39
3.3.6 P hotolum inescence ......................................................................... ..... 39
3 .3 .7 H all E effect .............................................................4 0
3.3.8 Current V oltage A nalysis..................................... ......... ............... 40
3.1.9 Capacitance V oltage A nalysis...................................... ............... 41
3.2 Diode Fabrication .................. ...... ............. .. ............. .. ............. 41
3.3 Proton Radiation Setup and Facility .............................................. ...............42

4 RADIATION AND THERMAL STABILITY OF VARIOUS TYPES OF
G A L L IU M N IT R ID E ......... .. ..................... ......... ..............................................53

4.1 Characterization of MOCVD N-type GaN.........................................................53
4.2 Characterization of MOCVD U-type GaN.........................................................57
4.3 Characterization of MBE Grown Ga-Polar GaN...............................................58
4.4 Characterization of MOCVD P-type GaN...........................................................58
4 .5 S u m m ary ...................................... ............................... ................ 5 9

5 RADIATION AND THERMAL STABILITY OF NOVEL OXIDES ON
G ALLIU M N ITRID E ............................................................. ............. ............... 106

5.1 Characterization of Sc203 on GaN ........................................ ............... 107
5.2 Characterization of Sc203/M gO on GaN ...........................................................108
5.3 Characterization of Sc203/M gCaO on GaN .................................. ........ ....... 109
5.4 Sum m ary ..................................... ................................ ......... 109

6 PROCESSING AND THERMAL STABILITY OF MOS DEVICES...................141

6.1 Processing of Novel Oxide/Gallium Nitride Devices .........................................141
6.2 Proton Radiation Effects of GaN MOS Diodes.............................................. 142
6.3 Effects of In-situ and Ex-situ Thermal Annealing.............................................143
6.4 Sum m ary .................................. ................................ ........... 146

7 SUMMARY AND DISSCUSSION .....................................................................172

7 .1 G aN v s. O x ide/G aN ............................................. ......................................... 172
7.2 M O S D devices ................................................ .. ...... ................. 173









L IST O F R E FE R E N C E S ......................................................................... ................... 176

BIOGRAPHICAL SKETCH ............................................................. ..................181
















LIST OF TABLES

Table p

2-1 Properties of Dielectrics Previously Studied for Use with GaN............................28

3-1 Proposed O xide Properties. ............................................................ .....................43

6-1 O xide Etchants and Condition........................................... .......... ............... 147

6-2 Dit Values at .4eV for Oxide/GaN Devices ..........................................................148
















LIST OF FIGURES


Figure page

2-1 Typical M O SFET................ ................. ........ .............. ............... .29

2-2 Cross-section illustration of a enhancement mode MOSFET.............................30

2-3 Cross section of a M O S capacitor .............................. .....................31

2-4 Dipole formation in the presence of an electric field .........................................31

2-5 Sketch of Riber MBE used for oxide growth (after Gila)............................... 32

2-6 AFM images of as received GaN. A) MOCVD GaN from Epitronics B) as
received M BE GaN from SVT. ........................................ ......................... 33

2-7 Examples of GaN surfaces before growth. A) The UV-ozone treated surface of
GaN. B) Buffered oxide etched surface of GaN .................................................34

2-8 Photos of RHEED indicating a (1x3) pattern. A) <11-20> crystal direction. B)
< 1-100> crystal direction ........................................................................ 34

3-1 Auger Electron Spectroscopy set-up ........................................ ...............44

3-2 Atomic force microscope (after K.K. Harris 2000). ............................................45

3-3 TEM setup used to image atomic layers at the film/substrate interface ...............46

3-4 SEM operation. Electron beam is rastered over the sample producing
secondary electrons (after S.M. Donovan 1999). .............................................47

3-5 Photolum inescence setup ......................................................... .............. 48

3-6 Ideal M O S I-V curve ................................................. ................................ 49

3-7 Ideal C-V curves of n-type MOS (after Johnson)...............................................49

3-8 Finished M OS capacitors design. ........................................ ....... ............... 50

3-9 Test Chamber for Radiation. A) Mounted samples in test chamber. B) Shutters
o f b eam .......................................................................... 5 1









3-10 Radiation Test Facility at Texas A&M University................................. 52

4-1 Initial PL scans of non-radiated 1st UOE n-GaN at temperatures 300K and
1 5 K ...................................... ................................................... . 6 1

4-2 Room Temperature PL analysis of 1st UOE n-GaN before and after radiation....62

4-3 15K PL analysis of 1st UOE n-GaN before and after radiation. ...........................63

4-4 Comparison of 300K 1st UOE n-GaN PL scans, pre-radiation, immediately
post radiation, and post radiation 1 year later........................................... ........... 64

4-5 PL analysis of annealed un-radiated 1 st UOE n-GaN. A) PL Spectra B)
B andedge inset. .......................................................................65

4-6 PL analysis of annealed 10MeV radiated 1st UOE n-GaN A) PL Spectra B)
B an dedg e in set (b elow )............................................................... .....................66

4-7 PL analysis of annealed 40MeV radiated 1 st UOE n-GaN. A) PL Spectra B)
B andedge inset. .......................................................................67

4-8 Normalized bandedge peaks of all annealed 1st UOE GaN to the 1st UOE GaN
control peak intensity of 1.4812 vs. the annealing temperatures. 0 temperature
indicates the initial state of the GaN .......................... ...... ............................. 68

4-9 Pre and post radiation PL analysis of 2nd UOE n-GaN ........................................69

4-10 PL analysis of un-radiated annealed 2nd UOE n-GaN. A) PL Spectra B)
B andedge inset. .......................................................................70

4-11 PL analysis of annealed 10MeV radiated UOE n-GaN. A) PL Spectra B)
Bandedge inset .................................... .......................................71

4-12 PL analysis of annealed 40MeV radiated UOE n-GaN. A) PL Spectra B)
B andedge in set ................................................................................... 72

4-13 Normalized bandedge peaks of annealed 2nd UOE n-GaN to the 2nd UOE n-
GaN control peak intensity of 12.354 vs. the annealing temperatures ................73

4-14 Pre and post radiation PL analysis of UF n-GaN .............................................74

4-15 PL analysis of annealed un-radiated UF n-GaN. A) PL Spectra B) Bandedge
in set. ................................................................................7 5

4-16 PL analysis of annealed 10MeV radiated UF n-GaN. A) PL Spectra B)
Bandedge inset .................................... .......................................76

4-17 PL analysis of annealed 40MeV radiated UF n-GaN. A) PL Spectra B)
B andedge in set ................................................................................... 77









4-18 Normalized bandedge peaks of annealed UF n-GaN to the UF n-GaN control
peak intensity of 25.069 vs. the annealing temperatures. .....................................78

4-19 Pre and Post Radiation of UOE u-GaN ...... ......... ...................................... 79

4-20 PL analysis of annealed un-radiated UOE u-GaN. A) PL Spectra B) Bandedge
in se t. ............................................................................ 8 0

4-21 PL analysis of annealed 10MeV radiated UOE u-GaN. A) PL Spectra B)
B andedge inset. .......................................................................81

4-22 PL analysis of annealed 40MeV radiated UOE u-GaN. A) PL spectra B)
B andedge inset. .......................................................................82

4-23 Normalized bandedge peaks of annealed UOE u-GaN to the UOE u-GaN
control peak intensity of 0.1162 vs. the annealing temperatures...........................83

4-24 Pre and post radiation PL analysis of UF u-GaN .............................................84

4-25 PL analysis of annealed un-radiated UF u-GaN. A) PL Spectra B) Bandedge
in se t. ............................................................................. 8 5

4-26 PL analysis of annealed 10MeV radiated UF u-GaN. A) PL Spectra B)
B andedge inset. .......................................................................86

4-27 PL analysis of annealed 40MeV radiated UF u-GaN. A) PL Spectra B)
B andedge inset. .......................................................................87

4-28 Normalized bandedge peaks of annealed UF u-GaN to the UF u-GaN control
peak intensity of 1.2997 vs. the annealing temperatures. ......................................88

4-29 Pre and Post Radiation PL analysis of MBE Ga-polar GaN...............................89

4-30 PL analysis of annealed un-radiated MBE Ga-polar GaN. A) PL Spectra B)
B andedge inset. .......................................................................90

4-31 PL analysis of annealed 10MeV radiated MBE GaN. A) PL Spectra B)
B andedge inset. .......................................................................91

4-32 PL analysis of annealed 40MeV radiated MBE GaN. A) PL Spectra B)
B andedge inset. .......................................................................92

4-33 Normalized bandedge peaks of annealed MBE Ga-polar GaN to the MBE Ga-
polar GaN control peak intensity of .01016 vs. the annealing temperatures.........93

4-34 Comparison of PL scans of as grown P-GaN and Activated P-GaN...................94

4-35 Pre and post radiation PL analysis of as-grown P-GaN.................. ...............95









4-36 Pre and post radiation PL analysis of Activated P-GaN.................. ...............96

4-37 PL analysis of annealed as-grown un-radiated P-GaN. ........................................97

4-38 PL analysis of annealed un-radiated Activated P-GaN ........................................98

4-39 PL analysis of annealed 10M eV as-grown P-GaN ................................................99

4-40 PL analysis of annealed 40MeV as-grown P-GaN. ........................... 100

4-41 PL analysis of annealed Activated P-GaN radiated at 10MeV............................101

4-42 PL analysis of annealed Activated P-GaN radiated at 40MeV............................102

4-43 Normalized bandedge peaks of annealed as grown P-GaN to the as grown P-
GaN control peak intensity of 2.7808 vs. the annealing temperatures ..............103

4-44 Normalized bandedge peaks of annealed Activated P-GaN to the Activated P-
GaN control peak intensity of .42224 vs. the annealing temperatures ..............104

4-45 All Normalized P-GaN bandedge peaks to control peaks of 2.7808 and .42224
respectively. ...................................................................... 105

5-1 Preliminary Sc203/GaN A) PL spectra and B) XRR scan ................................111

5-2 Sc203/GaN PL analysis. A) PL at 300K. B) PL at 15K..................................... 112

5-3 Sc203/GaN XRR scans showing the change in oscillations. ............................113

5-4 Sc203/GaN PL scans of pre-radiation, immediate post radiation and post
irradiation 1 year later. A) PL spectra B) Bandedge inset.............................114

5-5 PL scans of samples annealed non-radiated Sc203/GaN to 900C. A) PL Spectra
B ) B an dedg e in set ............................................................... .. ............. 115

5-6 Sc203/GaN post radiation anneals to 900C for radiation energies of 10MeV.
A) PL spectra B) Bandedge inset ........ .......................................... 116

5-7 Sc203/GaN post radiation anneals to 900C for radiation energies of 40MeV.
A) PL Spectra B) Bandedge inset.............................. ...............117

5-8 Sc203/GaN XRR scans of unradiated annealed sample at 400C. ....................... 118

5-9 Sc203/GaN XRR scans pre and post radiation annealed at 400C...................119

5-10 Normalized bandedge peaks of annealed Sc203/GaN to the Sc203/GaN control
peak intensity of .5139 vs. annealing temperature ...............................................120









5-11 Preliminary Sc203(20nm)/MgO(20nm)/GaN PL spectra and XRR. A) PL
spectra B ) X R R scan ......... .. ...... .... ..... ...... .................. ....................12 1

5-12 Sc203(20nm)/MgO(20nm)/GaN PL scans. A) PL at 300K B) PL at 15K.........122

5-13 Sc203(20nm)/MgO(20nm)/GaN XRR scan showing the change in oscillations
at 40 M eV ........................................................................123

5-14 Sc203(20nm)/MgO(20nm)/GaN PL scans of pre-radiation, immediate post
radiation and post irradiation 1 year later. A) PL spectra B) Bandedge inset. ....124

5-15 PL scans of samples annealed non-radiated Sc203(20nm)/MgO(20nm)/GaN to
900C. A) P1 Spectra B) Bandedge inset................................. .................. .....125

5-16 Sc203(20nm)/MgO(20nm)/GaN post radiation anneals to 900C for radiation
energies of 10MeV. A) PL spectra B) Bandedge inset............. ...............126

5-17 Sc203(20nm)/MgO(20nm)/GaN post radiation anneals to 900C for radiation
energies of 40MeV. A) PL spectra B) Bandedge inset.................................... 127

5-18 Sc203(20nm)/MgO(20nm)/GaN XRR scan of annealed sample. .....................128

5-19 Sc203(20nm)/MgO(20nm)/GaN XRR scans pre and post radiation annealed at
4 0 0 C ....................................................................... 1 2 9

5-20 Normalized bandedge peaks of annealed Sc203(20nm)/MgO(20nm)/GaN to
the Sc203(20nm)/MgO(20nm)/GaN control peak intensity of .21556 vs.
annealing tem perature ......... .............................................................. .... ..... 130

5-21 Preliminary Sc203(20nm)/MgCaO(20nm)/GaN. A) PL spectra B) XRR scan... 131

5-22 Sc203(20nm)/MgCaO(20nm)/GaN PL scans. A) PL at 300K B) PL at 15K
(b otto m ) ...................................... ............................................... 13 2

5-23 Sc203(20nm)/MgCaO(20nm)/GaN XRR scan showing the change in
o scillation s. ..................................................................... 13 3

5-24 Sc203(20nm)/MgCaO(20nm)/GaN PL scans of pre-radiation, immediate post
radiation and post irradiation 1 year later. A) PL spectra B) Bandedge inset. ....134

5-25 PL scans of samples annealed non-radiated Sc203(20nm)/MgCaO(20nm)/GaN
to 900C. A) P1 Spectra B) Bandedge inset.................................. ... ..................135

5-26 Sc203(20nm)/MgCaO(20nm)/GaN post radiation anneals to 9000C for
radiation energies of 10MeV. A) PL spectra B) Bandedge inset....................136

5-27 Sc203(20nm)/MgCaO(20nm)/GaN post radiation anneals to 9000C for
radiation energies of 40MeV. A) PL spectra B) Bandedge inset......................... 137









5-28 Sc203(20nm)/MgCaO(20nm)/GaN XRR scan of annealed sample.................... 138

5-29 Sc203(20nm)/MgCaO(20nm)/GaN XRR scans pre and post radiation annealed
at 4 0 0 C ..................................................................... 1 3 9

5-30 Normalized bandedge peaks of annealed Sc203(20nm)/MgCaO(20nm)/GaN to
the Sc203(20nm)/MgCaO(20nm)/GaN control peak intensity of 1.2968 vs.
annealing temperature ............ ............. .... ...................... 140

6-1 Finished diode m ask design. ......... ............................................ ............... 149

6-2 Ohmic IV plots taken from samples receiving various treatments....................150

6-3 Current density -Voltage behavior of Sc203/GaN after 10MeV irradiation. ......151

6-4 Capacitance -Voltage plot of Sc203/GaN after 10MeV irradiation ..................152

6-5 Current density -Voltage behavior of Sc203/ MgO/GaN after 10MeV
irra d ia tio n ...................................... ............. ............... ................ 1 5 3

6-6 Capacitance -Voltage plot of Sc203/ MgO/GaN after 10MeV irradiation..........154

6-7 Comparison of Current Density-Voltage results of different
Sc203/M gCaO/GaN thickness. ........................................ ........................ 155

6-8 Capacitance-Voltage plot of Sc203/MgCaO/GaN after 10MeV irradiation........156

6-9 Capacitance-Voltage plot of Sc203/MgCaO/GaN after 10MeV irradiation........157

6-10 Dit of post 10MeV radiation. A) Sc203/GaN B) Sc203/MgO/GaN.....................158

6-11 Current Density -Voltage behavior of Sc203/GaN during and after thermal
annealing ................................................................ .... ......... 159

6-12 Capacitance -Voltage plot of Sc203/GaN during and after heating ..................160

6-13 Current Density -Voltage behavior of Sc203/MgO/GaN during and after
heating..................................... ........................... ..... ........... 161

6-14 Capacitance -Voltage breakdown of Sc203/MgO/GaN after heating ................162

6-15 Current Density -Voltage behavior of Sc203/5nm MgCaO/GaN during and
after heating ..................................................................... 163

6-16 Capacitance-Voltage plot of Sc203/MgCaO/GaN during and after heating....... 164

6-17 Capacitance-Voltage plot of Sc203/MgCaO/GaN during and after heating. A)
CV of in-situ testing. B) CV of samples cooled to room temperature...............165









6-18 Capacitance -Voltage plot of Sc203/GaN after forming gas anneal at 400C.....166

6-19 Capacitance -Voltage plot of Sc203/MgO/GaN after forming gas anneal at
4 0 0 C ....................................................................... 1 6 7

6-20 Capacitance-Voltage plot of Sc203/MgCaO/GaN after forming gas anneal at
4 0 0 C ....................................................................... 1 6 8

6-21 Dit of Annealed Sc203/GaN. A) Probe station thermal anneals. B) Forming
G as anneal............................................................................................. 169

6-22 Dit of Annealed 2nm Sc203/40nm MgO/GaN. A) Probe station thermal
anneals. B) Form ing gas anneal .............. ........................................................ 170

6-23 Dit of Annealed 35nm Sc203/5 nm MgCaO/GaN A) Probe station thermal
anneals. B) Form ing gas anneal .............. ........................................................ 171

7-1 Recovery of Oxide/GaN post 9000C after 7 months.................. .. ..................175















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

PROTON RADIATION AND THERMAL STABILITY
OF GALLIUM NITRIDE AND GALLIUM NITRIDE DEVICES

By

Kimberly K. Allums

May 2006

Chair: Cammy R. Abernathy
Major Department: Materials Science and Engineering

In today's industry one can see a constant challenge to exceed the limits of

yesterday's devices. For the last three decades, the III-V nitride semiconductors have

been viewed as highly promising for semiconductor device applications. The primary

focus of III-V nitrides, thus far, has been centered on light emitting diodes (LEDs),

injection lasers for digital data reading and storage applications, and ultra violet

photodetectors. Yet, another application is high-power electronic devices for space-borne

communications systems. It is expected that GaN-based devices will be more resistant to

radiation damage often encountered in space environments, though verification of this is

just now being undertaken. In particular, no information is yet available about the

sensitivity to radiation of devices using dielectrics such as MOSFETs. Similarly, very

limited data has been reported on the effects of high-energy protons on GaN based

devices of any type. For this reason the research presented in this dissertation was

undertaken to study the radiation and thermal stability of gallium nitride materials and









gallium nitride semiconductor diodes, with and without novel gate dielectrics such as,

scandium oxide (Sc203) and magnesium oxide (MgO) and the ternary mix of magnesium

calcium oxide (MgCaO).

It was found that though environmental degradation could be a problem for MgO

dielectrics, the radiation exposure itself did not produce significant damage in either the

Sc203, MgO or MgCaO dielectrics. Much of the minimal damage occurred in the GaN as

shown by photoluminescence spectroscopy (PL).














CHAPTER 1
INTRODUCTION

In today's industry, microelectronics has been largely based on silicon solid-state

devices. As technology has improved, the demand for devices that can operate at higher

temperatures and in more caustic environments has become the focus of several research

areas especially military applications. As a result, compound semiconductors are

becoming increasingly important because they do not possess the limitations of Si-

devices and have direct bandgaps that allow them to be used in optical applications.

Much of the focus of compound semiconductors has been on the III-V groups.

Compound semiconductors, such as gallium arsenide (GaAs) and indium phosphide

(InP), have higher carrier mobilities and lower-saturation electric fields than silicon

semiconductor devices, and research based on these compound semiconductors have led

to breakthroughs in device performance. For the last few decades the III-V nitrides, such

as gallium nitride (GaN), have been viewed as highly promising for semiconductor

device applications mainly because of a wider bandgap that will allow the semiconductor

to overcome thermal and power handling limits of the GaAs and InP semiconductors. The

primary focus of GaN applications thus far has centered on light emitting diodes (LEDs),

injection lasers for digital data storage and ultraviolet photodetectors. Because of its large

bandgap, Eg, GaN can operate at higher temperatures than its other semiconductor

counterparts. However, the large Eg also requires a large bandgap oxide in order to

provide adequate carrier confinement in MOS-type applications. Various oxides were

tested in previous studies but have been proven to possess many limitations as gate









dielectrics. Some alternative materials that are candidates for an optimum dielectric for

GaN are magnesium oxide (MgO), and a ternary combination of magnesium calcium

oxide (MgCaO). For use in wide bandgap semiconductor devices, the dielectric materials

must also possess excellent thermal stability, both because of the high operating

temperatures and the high processing temperatures needed for device fabrication. These

oxides must also be radiation resistant for use in low orbit and aerospace applications.

The ultimate goal of this project is to provide a high quality oxide that can improve the

radiation hardness and thermal stability of GaN devices.














CHAPTER 2
BACKGROUND AND LITERATURE REVIEW

2.1 Introduction

This chapter discusses the fundamentals of field effect transistors (FET), metal

oxide semiconductor capacitors, growth and characterization of gallium nitride, dielectric

films and their properties, oxide growth processes and device applications. Also included

is a literature review of the possible effects of radiation and the characterization methods

used to determine the stability of the heterostructures in question.

2.2 Device Structures

A field effect transistor (FET) is a unipolar device where only one type of carrier

takes part in the conduction process. The FET is a three-terminal device in which the

current through two terminals is controlled by a voltage applied at the third. FETs are

characterized by high input impedance since the control voltage is applied to a reverse

biased junction either through a metal Schottky barrier or across an insulator [1]. Some

advantages of FETs include higher switching speeds and higher cutoff frequencies than

bipolar devices. There are various types of FETs such as junction field effect transistors

(JFET) and metal semiconductor field effect transistors (MESFET), but the focus of this

study will be metal oxide semiconductor field effect transistors (MOSFET) and metal

oxide semiconductor (MOS) capacitors.

2.2.1 Metal Oxide Semiconductor Field Effect Transistor

The MOSFET is one of the most important devices for very large-scale integrated

circuits such as microprocessors, and semiconductor memories and is also becoming an









important power device [2]. Similar devices to the MOSFET include the metal insulator

semiconductor field effect transistor (MISFET) and the insulated gate field effect

transistor (IGFET). For all of these device types, the gate region of the transistor

determines the capabilities of the device. The uniqueness of the MOSFET relies on the

oxide layer's ability to prevent current flow from the semiconductor to the gate due to the

high resistivity of the oxide. Figure 2-1 shows a general structure of a MOSFET. Ideally

a MOSFET should possess high output current drive, ID, high transconductance, gm,

stable threshold voltage, VT, fast switching speed, high gate oxide breakdown voltage and

low source/drain to body capacitance. Physically, the current drive is linearly related to

gate width and as ID increases so does the capacitance.

There are two types of MOSFET devices, depletion mode and enhancement mode.

In the depletion mode device, the material under the gate is doped in order to carry

current. This device is in the "ON" state when there is no applied gate voltage. When

there is a zero gate voltage, carriers are free to flow from the source to the drain in the

MOSFET structure. As a negative voltage is applied to the gate contact, the area under

the gate, called the channel, is gradually depleted of carriers. The depletion depth is

increased until the flow from source to drain is stopped. This voltage is called the pinch

off voltage, since it effectively pinches the channel shut. The transistor is now "OFF."

As the voltage across the source-drain is increased, it requires more gate voltage to

successfully pinch-off the carrier flow. In the enhancement mode device, the material

type under the gate is not doped; thus no channel is present and in an OFF state since

there is no current flow. As with the depletion mode MOSFET, a positive drain voltage is

still applied but there is also a positive gate voltage applied as well. This has the effect of









attracting free electrons towards the gate, thus creating a channel of free flowing

electrons and the larger the positive gate voltage the wider the carrier channel of

electrons. Figure 2-2 shows an example of the ON and OFF states of the enhancement

mode device. Thus the maximum operating parameters of the device are determined by

the amount of electric field that can be applied to the gate before dielectric breakdown

occurs. Also, forward gate voltages can be used to increase the amount of current, which

can be passed through the channel. Again a high dielectric breakdown field is required.

The MOSFET has several advantages over heterojunction type transistors, such as

relative insensitivity to temperature during operation. Also MOSFET devices are

expected to have a wide gate modulation range. This wide range is because the device

turn-on is dependent on the dielectric thickness and is not limited to low turn-on voltages

like those obtained when using Schottky metal contacts.

2.2.2 Metal Oxide Semiconductor Capacitors

The MOS capacitor (MOSC) is the most basic device used to evaluate the electrical

properties of the oxide and the semiconductor. Throughout this study the MOS capacitor,

also known as the MOS diode, is used for evaluation of the MOS structure. For use with

GaN on sapphire substrate, the contacts are created front side. The process is done by

etching the oxide to expose the underlying semiconductor material and then depositing

ohmic and gate contacts. Figure 2-3 shows a cross section of the MOSC. The MOSC is

operated by applying a potential difference across the gate and ohmic contacts. An

understanding of MOSC behavior is obtained through the use of energy band diagrams

and ideal capacitance-voltage curves.









2.3 Growth and Photoluminescence Spectroscopy of Gallium Nitride

One of the ways high quality gallium nitride is grown is via metal organic chemical

vapor deposition (MOCVD). Trimethlgallium (TMG) and ammonia (NH3) are used as

source gases for Group III and V species to obtain c-axis oriented films of GaN grown on

(0001) sapphire (A1203) wafers. In MOCVD the flow of the reactant gas sources is very

critical to the GaN film growth and any slight change can alter the film quality. Hydrogen

is normally used as a carrier gas and is flowed normal to the substrate surface to bring the

reactant gases in contact with the substrate and to prevent thermal convection effects.

MOCVD can be used to grow n-type and p-type GaN. Undoped GaN is naturally n-type

due to nitrogen vacancies that occur in the film. Two types of GaN are generally

mentioned when referring to n-type GaN, doped and unintentially doped. Silicon is

normally used to doped GaN to increase the number of electron carriers within the

substrate and the carrier concentration is typically -1019cm-3 whereas unintentially doped

GaN (known as u-GaN) has a typical intrinsic electron carrier concentration of ~1016 cm

3. As well as being n-type, MOCVD GaN is normally grown with N terminated faces at

the surface which tend to have dangling bonds that allow for the surface to be susceptible

to contamination and impurities. This can lead to surface effects such as surface charging

and fermi level pinning. P- type GaN is made by doping the GaN with magnesium, Mg,

to create majority carriers of holes.

Gallium nitride can also be grown via molecular beam epitaxy (MBE) with either N

terminated ([0001] direction) or Ga terminated ([000-1] direction) surfaces using sapphire

substrates. MBE's inherent control over growth parameters can be used to interrogate

certain structural and electrical processes in the crystal. It has been successfully shown

that GaN can be grown with Ga terminated surfaces (known as Ga-polar GaN), which









results in a smoother surface morphology and better quality films. The successful growth

of Ga-polar GaN via MBE has a very narrow margin of error when it comes to the exact

growth parameters and to assist with the correct termination many growths of Ga-polar

GaN start with a MOCVD GaN template, which has the right orientation for continued

growth of the GaN film by MBE.

Defects in GaN degrade device performance and longevity. Therefore it is

important to have an understanding of the types of defects present and the effects they

produce in GaN. Native defects in GaN tend to be Ga or N vacancies, impurities,

dislocations and/or interstitial atoms. One non-destructive characterization used to

determine the types of defects and their effects is photoluminescence spectroscopy (PL).

These defects cause different energy or color bands in the PL spectra. By monitoring the

intensity changes and wavelength peak shifts in these bands, it is possible to identify the

types of defects that are present. Knowing this information concerning the defects can

allow for treatments pre or post GaN growth to reduce the defects. Common color bands

that show up in GaN are the ultraviolet band (UVL) at approximately 390-455nm

(3.076eV- 3.257eV), blue band (BL) at approximately at 438 to 442nm (2.88eV -

2.90eV), green band (GL) at 492 to 577nm(2.5eV 2.2eV), the much debated yellow

band (YL) at 577 597nm (2.2eV 2.3eV), and a red band (RL) at 688nm 652nm

(1.8eV- 1.9eV).

Each of these bands corresponds to an energy transition within the band gap of

gallium nitride and depending on the energy or band associated with the transition,

pertains to a particular type of defect within the material. The ultraviolet band (UVL) has

shown up in undoped GaN as well as Mg-doped GaN. In u-GaN the main candidates for









shallow donors are SiGa and ON. The assignment of shallow acceptor has been identified

using optical detected magnetic-resonance (ODMR) as SiN and in Mg-doped GaN, the

shallow acceptors are MgGa and SiN. The UVL bands in each type of GaN are very similar

to each other indicating the different shallow acceptors manifest themselves similarly in

GaN as a transition from a shallow donor to a shallow acceptor.

The blue band that occurs approximately 420nm to 490nm is observed in undoped,

and Mg-doped GaN but only those grown via MOCVD or HVPE because the defect is

not a native defect that is grown into the GaN film. The BL band has been defined in u-

GaN as a transition from a shallow donor to deep acceptors at low temperatures and at

elevated temperatures from conduction band to deep acceptor. In p-GaN the transition

was described as a transition from deep donor to the shallow MgGa acceptor. Quenching

of the BL band occurs when holes from the acceptor level escape to the valence band and

an observation of the BL band at low temperatures such as 15K enables the determination

of the vibrational characteristics of the defects in its ground state.

The green band in GaN is usually coupled with the YL band and sometimes

overshadowed by the intensity of the YL band. However, in high purity GaN a yellow-

green band can be observed at room temperature. The shape and position of the band

depends on the excitation intensity and excitation energy and can always be deconvolved

into two bands. It is presumed that the YL and GL band are related to two charge states

of the same defect, presumably the VGaON complex.

Many researchers have disputed the attribution of the YL band in GaN and are still

undecided as to the exact orgins of the YL band. Although they have agreed that the YL

band is not specific to a particular impurity defect, they have instead related the YL band









to a native defect that is grown into the gallium nitride and is specific to a type of

transition with the band gap. It is also believed that in undoped GaN the VGaON complex

is responsible for the YL band and that ON and CGa are shallow donors. It is still debated

whether or not the YL band is a transition from the conduction band or shallow donor to

a deep acceptor or as one researcher suggested [hoffman], a two stage process that

involves a non-radiative capture of an electron by a deep double donor followed by a

radiative recombination between the electron at the deep donor and a hole at a shallow

acceptor. Whichever transition is responsible, thus far no model fully describes the defect

or defects responsible for the YL band. What has been found about the YL band is that

carbon doping and Si doping to a certain concentration enhance the YL emission. It is

believed that the carbon and silicon sit upon Ga vacancies. For Si, the initial enhancement

comes from inadvertent increase in the concentration of the VGa-related acceptors such as

VGaON and VGaSiGa facilitated by the shift of the Fermi level to higher energies during

growth. The activation energy of the SiGa donor is 30meV. Although carbon does

enhance the YL emission, the likelihood of a CGa formation is very low.

The red band in u-GaN is the least studied and the least common of all the color

bands that appear. The intensities are independent of temperature, saturate at low

excitation, and possess long lifetimes. The transition is known to be shallow donor or

conduction band to a deep acceptor at low temperatures and elevated temperatures

respectively.

In Ga-polar GaN two color bands dominate within the low temperature PL spectra,

a red band (RL2) and a green band (GL2). Both tend to have very weak emission and

quench above 100K. These bands are not to be confused with the other RL and GL bands









mentioned. The bands are created by point defects that are fairly uniform throughout the

bulk GaN and tend to dominate high-resistivity GaN (no=1015 cm-3). Due to internal

transitions in the defects, the Gaussian shape of the bands indicate that in both cases the

carrier is strongly localized at the defect providing a strong electron-phonon coupling.

In summary, determining the color spectra of PL is a very useful way to thoroughly

characterize all types of gallium nitride films and to determine what defects are grown in

and what defects are induced due to doping, impurities, and irradiation.

2.4 Dielectrics

2.4.1 Ideal Dielectrics

Ideal dielectric materials are perfect insulators in which no mobile charged particles

are present. The ability to store an electric charge is called capacitance [3]. When an

insulator is placed between the conducting plates of a capacitor, the capacitance can

increase significantly. Insulators have positive and negative charges in the form of the

atom nucleus and the electron cloud, but these charges are bound to the atom or molecule

and are not available for conduction [4]. Under the influence of an electric field, the

nucleus and the electron cloud are displaced to a form a dipole [1]. Figure 2-4 shows the

dipole formation in the presence of an electric field. The total effect of an electric field on

a dielectric material is known as polarization. The ability of the material to resist the

polarization of charge is described as the dielectric constant, K, which is the ratio of the

permittivity of the material, si, to the permittivity of vacuum, ev.

K=Si/Sv (2-1)

The dielectric constant can also be related to the internal field created within the

material and the external field applied, through Equation 2-2 [5].









Einternal = Eapled /K (2-2)

The polarization, P, of the material is related to the dielectric constant by equation

2-3, where is the strength of the electric field (V/m). It can be assumed from this

relation that the polarization increases as the electric field strength increases, until all the

dipoles are aligned such that

P=(K-1);o, (2-3)

There are various applications for dielectric materials. Passivation of high voltage

junctions, isolation of devices and interconnects and gate insulation of field effect

transistors are a few that are relevant for this discussion. For a material to be a successful

dielectric it must meet certain criteria. Desirable characteristics include chemical

stability over the life of the device, immobile charge traps (to avoid shorting and

frequency limits) and a dielectric constant higher than that of the semiconductor (to avoid

generating a high electric field in the dielectric). The dielectric/semiconductor interface

is also an important focus of research in the area of device processing. The interface state

density of carrier traps must be
successful. A good dielectric should possess low defect density, good barrier properties

against impurity diffusion, high-quality interface with low state density and fixed charge

and stability under irradiation. For use in wide bandgap semiconductor devices, the

dielectric materials must also possess excellent thermal stability, since use at high

operating temperatures is very desirable.

2.4.2 Crystalline vs. Amorphous

There are several reasons that both crystalline and amorphous dielectric layers are

being investigated. The major concerns with MOS-type semiconductor devices are the









interface between the dielectric and substrate and current leakage in the oxide layer.

Ideally the perfect dielectric would be a single crystal structure with the same symmetry

and no lattice mismatch to GaN. The absence of lattice mismatch eliminates stress and

defects at the interface. Realistically, using a dielectric material with small lattice

mismatch to the substrate should reduce high defect densities at the interface. However,

in some cases it may not be possible to eliminate a sufficient number of crystalline

defects; thus the possibilities of using amorphous oxides will also be discussed.

Using a single crystal layer of the prospect oxides should in principle provide a

high quality layer. However, even the small lattice mismatch of the two prospect oxides

can still cause defects in the interface. These defects can propagate through the entire

oxide creating traps or leakage paths. One problem is finding the right growth conditions

that will produce a single-crystal and not a poly-crystalline structure. Polycrystalline

layers at the interface could result in grain boundaries allowing high leakage current

through the dielectric. In an amorphous arrangement the atoms are randomly distributed

on the substrate. This eliminates the problem of strain and dislocations at the interface

and in the dielectric. A truly amorphous layer would also stop any shortening of current

through the oxide, because the electrons or holes would have no defect paths to follow.

One potential problem is the possibility for formation of crystallites in the layer. At high

operating temperatures, the crystallites may grow and coalesce to form grain boundaries

and a polycrystalline-like structure, resulting in a very leaky and thermally unstable

dielectric. Each type of oxide layer has advantages and disadvantages.

2.4.3 Present State of Dielectrics for Gallium Nitride

Several dielectrics have been tested for use on wide bandgap GaN semiconductor

devices (Table 2-1). Gallium nitride research has resulted in long-lifetime, room









temperature operation of photonic devices. These include LEDs that cover the visible

spectrum, laser diodes in the blue and blue-green regime, and UV detectors. These

devices are just recently reaching production with problems still to be solved in the fields

of n-Ohmic and p-Ohmic contacts, p-type doping issues, Schottky contacts, and dielectric

materials. Also, with the lack of availability of high-quality GaN substrates, research an

epitaxial growth and substrate selection is still ongoing. From material and processing

advances learned from the photonics research, high-power and high-temperature switches

have been realized. The following summarizes some of the dielectrics research to date.

2.4.3.1 Silicon oxide on GaN

Silicon oxide is a very attractive choice for a dielectric material since it is has been

well studied and the processing is well established. Silicon oxide deposited by plasma

enhanced chemical vapor deposition (PECVD) [6-9] has been reported to give interface

state densities on the order of low 1011 eV-cm-2. Silicon oxide deposited by electron

beam (EB) evaporation on GaN has shown an interface state density of 5.3x101 eV cm-2

[8]. After annealing the EB deposited SiOx at 650C, the valance band offset was

measured to be 2eV and the conduction band offset was measured to be 3.6 eV [10]. The

EB evaporated SiOx shows a silicon rich stoichiometry when compared to the PECVD

SiOx.

There are several inherent problems with SiOx as a dielectric material for wide

bandgap semiconductors. The high Dit of SiOx is due to uncontrolled oxidation of the

surface [11]. The most significant limitation is that SiOx has a dielectric constant (E) of

3.9, which is significantly lower than that of GaN (E = 8.9). This will create a very large









electric field in the dielectric, leading to further breakdown and increased power

consumption.

2.4.3.2 Silicon nitride on GaN

The Dit obtained from PECVD silicon nitride [8,9] has been reported as 6.5x1011

eV-cm-2. The Si3N4 MISFET ,however shows poor performance. Electrical

measurements showed the MISFET structure had a large flat band voltage shift (3.07 V)

and a low breakdown voltage (1.5 MV/cm). There was no microstructural analysis

performed on the deposited Si3N4 films. SiN deposited by ECR plasma CVD showed a

Dit of lxl10lcm-2eV- but had excess leakage current due to the small conduction band

offset [11]. The ECR-CVD MIS diode showed a Dit of 4xl011 cm-2eV-1, fixed oxide

charge of 1.1x1011 cm-2 and breakdown 5.7 MV/cm with a dielectric constant of 6 [12].

A unique dielectric structure of SiO2/Si3N4/SiO2 (ONO) was reported to have

breakdown field strength of 12.5 MV/cm for temperatures as high as 3000C.20 The ONO

structure was deposited by jet vapor deposition to a thickness of 10 nm /20 nm /10 nm.

The multiplayer structure does allow for unique engineering of a dielectric, but multiple

interfaces can lead to an extremely large number of interface state traps and increased

processing. The Dit for the ONO structure was reported to be less than 5x101 eV cm-2

with breakdown fields greater than 12 MV/cm [13].

2.4.3.3 Aluminum nitride on GaN

Aluminum nitride deposited by MBE and MOCVD has been used to create

MISFET devices and insulated gate heterostructure field effect transistor (IG-HFET)

devices [14,15]. The A1N MISFET structure grown at 4000C was polycrystalline. From

x-ray reflectivity measurements, the AIN/GaN interface showed a roughness of 2.0 nm.









This may be due to the polycrystalline nature of the film or from intermixing of the A1N

and GaN at the interface. The dielectric breakdown field was calculated to be

1.2 MV/cm. The A1N IG-HFET structure was grown at 9900C, forming a single crystal

film of 4.0 nm. This device operated in depletion mode and had a pinch-off voltage of 0

V. Hexagonal aluminum nitride has a 2.4% lattice mismatch with hexagonal GaN on the

(0001) plane. The 4.0 nm film thickness is greater than the critical thickness allowed for

elastic deformation leading to threading dislocations forming from plastic deformation.

Single crystal A1N and polycrystalline A1N films suffer from defects and grain

boundaries that cause shorting.

2.4.3.4 Gallium oxide on GaN

Like Si, gallium nitride forms a native oxide. This oxide has been considered as a

dielectric material, like the native oxide of silicon. Thermal oxidation has been studied

using dry [16,17] and wet [18] atmospheres. Dry oxidation of GaN epilayers at

temperatures below 9000C showed minimal oxidation. Dry oxidation at 8800C for 5

hours produced 1110 nm of P-Ga203 with a Dit of lx1010 eV-cm-2 and showed inversion

[19]. At temperatures above 900C, a polycrystalline monoclinic Ga203 formed at a rate

of 5.0 nm/hr. This oxidation rate is too slow to be viable as a processing step. Wet

oxidation of GaN also forms polycrystalline monoclinic Ga203, but at a rate of 50.0

nm/hr at 9000C. From cross-sectional transmission electron microscopy, the interface

between the oxide and the GaN was found to be non-uniform. Scanning electron

microscopy showed that both films were rough and faceted. Electrical characterization of

the oxide showed a dry oxide dielectric field strength of 0.2 MV/cm and a wet oxide

dielectric field strength of 0.05 to 0.1 MV/cm. The microstructure formed from this









process was shown by XRD to be a high temperature hexagonal phase [20]. Ga203,

formed by this method passivates the surface [20-23] and has a Dit of 1011 eV-cm-2 for

GaN MOS. A negative oxide charge as well as high capacitance and reduced reverse

leakage where shown for thicker oxides grown by PEC [21]. Using PEC and a HeCd

laser, a low reverse leakage current of 200 pA at 20Vm was achieved. For this oxide the

forward breakdown, Efb, was 2.8 MV/cm and the reverse breakdown, Erb, was 5.7 MV/cm

with a Dit of 2.53 x 1011 cm-2eV-1. The dielectric constant of Ga203 grown under these

conditions was estimated to be10.6 [24]. With amorphous GaO deposited by PEC, low

leakage currents of <10 x 10-8 A/cm at -15 V have been measured. Using Ga203 as both

gate dielectric and passivation layer, a breakdown field of 0.4 MV/cm was observed. The

bandgap, Eg, of Ga203 was measured to be 4.4 eV [25].

2.4.3.5 Silicon dioxide on gallium oxide on GaN

Depending on the growth technique, the interface between the SiO and the GaN can

vary drastically. When SiO is deposited by RPECVD, a parasitic subcutaneous layer of

native gallium oxide is grown on the GaN surface. This layer has been shown to have a

direct effect on the device performance. When the thickness of the initial GaO layer is

controlled by a pre-oxidation step the device characteristics improve markedly. Remote

Plasma Assisted Oxidation first followed by RPECVD gave a lower Dit and a smaller flat

band shift over the RPECVD of the SiOx alone [26]. Real and ideal CV curves are nearly

identical [27]. After an anneal in an RTA for 1 min at 9000C in Ar the Dit is

2-3x1011 cm-2 [28,29]. Another group used a similar oxide growth technique and

measured a Dit of 3.9x1010 eV- cm2 and a low leakage current [30, 31].









2.4.3.6 Gallium gadolinium oxide on GaN

Due to the recent success of gallium gadolinium oxide (GGG) as a dielectric in

GaAs MOSFETs [32-37], attention has turned toward this as a dielectric material for

GaN. The GGG dielectric was deposited on a GaN epilayer by EB evaporation of a

single crystal GGG source [14]. The substrate temperature was 550C. The interface

roughness was calculated to be 0.3 nm from x-ray reflectivity. Metal oxide

semiconductor (MOS) capacitors were formed and tested. A breakdown field of greater

than 12 MV/cm was estimated. More recently, the thermal stability of the film and the

interface has been proven to temperatures as high as 950C and operation of a depletion

mode MOSFET has been performed at temperatures up to 4000C [38]. The EB

evaporated GGG stoichiometry is heavily dependant upon the substrate temperature.

Changes in temperature lead to changes in the stoichiometry [39].This limits the available

microstructure obtainable within the stoichiometric limits of GGG.

2.4.3.7 Gadolinium oxide on GaN

GaN based MOSFETs have been made that used a stacked gate oxide consisting of

single crystal gadolinium oxide and amorphous SiO2 [40]. The gadolinium oxide

provides a good oxide /semiconductor interface and the SiO2 reduces the gate leakage

current and enhances oxide breakdown voltages. The dislocations in the Gd203 film limit

the breakdown field that can be sustained in the dielectric.

2.4.3.8 Scandium oxide on GaN

Scandium oxide grown by MBE has been used as a gate dielectric and passivation

layer for GaN based devices. Scandium oxide has the bixbyite crystal structure, a

reasonable band gap of 6.3 eV and a lattice mismatch to GaN of 9.2%. The scandium









oxide was grown by MBE using an RF oxygen plasma, substrate temperature of 6500C

and an effusion cell temperature of 11300C. This oxide showed inversion when used in a

GaN gated diode [41]. This oxide was also grown under the same conditions except the

substrate temperature was 1000C, which resulted in an interface state density of 5xl011

eV- cm2 [42].

Scandium oxide has also been used as a field passivation layer for GaN devices. It

has been shown to reduce the reverse leakage current and increase the fT and fMAX. The

passivation films have been grown with a substrate temperature of 1000C and a cell

temperature of 1130C [43]. Scandium oxide has better long-term stability than SiNx as a

passivation film for GaN based HEMTs. It has been shown to dramatically reduce the

gate lag problems due to surface states on AlGaN/GaN HEMTs [44].

2.5 Epitaxal Growth

The term "epitaxy is Greek in origin, coming from the words "epi" meaning

upon, and "taxis" meaning to arrange. Epitaxy is the process of controlled growth of a

high-quality crystalline or amorphous layer of material on a substrate. There are two

categories of growth for epitaxy: homoepitaxy and heteroepitaxy. Homoepitaxy is the

growth of a crystalline layer that is the same as the substrate material while heteroepitaxy

is the growth of a layer that differs from that of the substrate [1]. The growth method

used in this study is Molecular Beam Epitaxy (MBE). It is important to select growth

parameters that will produce the highest quality epitaxial layer for the GaN

semiconductor, in order to reduce crystalline defects that would degrade the performance

of the semiconductor device.









2.5.1 Molecular Beam Epitaxy

The growth of the epitaxy layer takes place due to reactions between the molecular

beams coming from the sources and a crystalline surface held at suitable temperatures.

The molecular beams are produced from solid element at sources heated in effusion cells.

Growth begins when different atomic species are absorbed on the substrate surface and

migrate to form the deposited layer. A common in-situ growth-monitoring technique is

electron diffraction, commonly known as RHEED (Reflection High-Energy Electron

Diffraction). In this method, high-energy electrons are diffracted off the growing surface

and produce an image on a screen on the opposite side of the chamber [45] (Figure 2-5).

Although MBE has the capability for growing complex multilayers where precise

control of dopant concentration, layer thickness and interface abruptness are required,

there are also serious concerns. For example, due to the UHV environment sensitivity to

contamination and replenishment of sources, long periods of system downtime can ensue.

2.5.2 Substrate Preparation

Before any epitaxial growth, the substrates received an ex-situ and an in-situ

surface treatment to remove any contamination and the native oxide. The surface of the

semiconductor must be as clean and planar as possible to ensure high quality dielectric

film deposition. Surface contamination leads to impurities at the

dielectric/semiconductor interface, which ultimately result in the formation of interface

traps. Since gallium nitride substrates are not currently available, gallium nitride grown

on sapphire wafers oriented <0001> were used. These will be referred to throughout this

work as GaN substrates. Two different types of growth of the GaN substrates were

employed in this work, MBE and metal-organic chemical vapor deposition (MOCVD).

The MBE GaN substrates were grown in house (referred to as UF MBE GaN) and the









MOCVD substrates were either grown in house (referred to as UF GaN) or by Uniroyal

Optoelectronics (UOE). From AFM, there is a large difference in surface roughness

between the two types of GaN. The MOCVD substrates are 1-3 nm RMS roughness and

the MBE substrates are -6 nm RMS roughness (Figure 2-6).

The GaN substrates received an ex-situ treatment consisting of a 3-minute etch in

(1:1) hydrochloric acid: water, followed by a DI rinse and blown dry by nitrogen. This

was use to remove any organic residue from the surface. Then a 25-minute exposure to

ozone produced by an ultraviolet lamp in a UVOCS UVO Cleaner model number 42-220

was used to oxidize the carbon on the surface and create a thin native oxide layer. Next,

the substrates received another etch in buffered oxide etch for 5 minutes to remove the

native oxide. This is shown by observing the RHEED pattern produced from the surface.

The RHEED pattern produced by the native oxide is more diffuse than the pattern

produced by the buffered oxide etched surface (Figure 2-7).

The GaN substrates were then mounted to molybdenum blocks using indium

solder, and then loaded into the load-lock of the MBE. Room temperature RHEED

measurements showed a reasonably clean (lxl) surface (Figure 2-7). Two different

crystal directions are observed in RHEED to create a more complete understanding of the

surface. Here, the <1-100> and the <11-20> directions are observed. An in-situ thermal

treatment was employed to further remove any oxide or contamination left on the surface.

The substrates were heated to 7000C in vacuum and no overpressure of nitrogen was

used. The RHEED patterns recorded at this temperature indicate a sharp (1x3) pattern

(Figure 2-8).









2.6 Types and Effects of Radiation on Microelectronics

In the past decades most radiation testing has been performed on Si devices.

Currently there is very limited information on radiation effects on GaN devices. The

objective is to investigate the stability of the oxide, the interface between the oxide and

semiconductor, and the operational characteristics of the MOS capacitor.

In order to understand the relevant applications for radiation hard devices one

must first understand the origins and effects of the space radiation and terrestrial radiation

environment. There are various types of radiation that affect electronic devices such as

gamma rays, x-rays, protons, neutrons and other subatomic radioactive particles. Much of

this radiation comes from solar cosmic rays, galactic cosmic rays that originate outside

the solar system, and other planetary environments. The earth's magnetic field shields a

region of near earth space from these particles but they easily reach polar regions and

high altitudes such as geosynchronous orbit (35,800km) in which satellites travel [46,47].

Around the earth lies a radiation environment known as the Van Allen belts. These belts

are compromised of trapped subatomic particles that can affect electronic devices

operating at high-altitudes.

Gamma rays are electromagnetic radiation emitted by radioactive decay. These

rays are shorter in wavelength than ultra violet and are far higher in energy. For

experimental use, common sources of gamma rays are Cobalt 60 and Cesium 137. Effects

of gamma rays tend to occur over long terms and can cause changes in electrical

operational characteristics.

X-rays are created from high-speed electrons that after impinging on a metal

release energy in the form of an electromagnetic wave. Unlike gamma rays, x-rays

consist of a mixture of different wavelengths. In addition to originating from solar flares









and other cosmic occurrences, x-rays are also produced by nuclear explosions. After a

nuclear explosion, much of the kinetic energy of the fission fragments is converted to

internal excitation and radiation. Some 70-80% of the total energy of this material is

emitted as thermal electromagnetic radiation in the soft x-ray region. This can produce

ionizing effects on MOS devices. Radiation of this type can be found at satellite orbit

altitudes as well.

Neutrons are also a concern with radiation environments. Neutrons are released

following nuclear fission and can travel large distances even at sea-level and at

atmospheric pressures. The neutrons can produce recoil protons by elastic collisions in

hydrogen rich materials, therefore ionizing the material or electronics of interest.

Each of these types of radiation can damage or upset electronic devices but in

comparison to the amount of protons in the environment they make up a small

percentage. Generally, protons trapped in the earth's magnetosphere have energies up to

800 MeV. They primarily originate from galactic cosmic rays, which come from outside

the solar system. These protons are concentrated in a small area known as the South

Atlantic Anomaly. The radiation belt dips into the earth's atmosphere due to the earth's

tilt on its axis. This causes concern for electronic devices that will operate in or near the

region. Hence the focus of this study, effects of proton radiation on gallium nitride

devices. Proton radiation causes ionization in electronic devices. This involves creating

electron-hole pairs, Frenkel defects, and bulk semiconductor defects [48]. Proton

radiation can also cause traps to form at the interface of the oxide and substrate. These

various radiation-induced defects can cause current leakage through the oxide,

displacement or removal of semiconductor atoms, build up of trapped charge and shifts in









operational parameters of the electronic device. Previous studies on silicon devices have

shown radiation resistance under the mentioned environments. Gallium nitride is

expected to be even more radiation resistant or rad-hard than silicon. In the course of this

study 40MeV(+- 2%) and 10MeV (+- 10%) energies of protons will be examined at

5x109cm2 total particle dose. Although lower energy protons can be blocked by shielding,

it is useful to use lower energy protons in order to look at the type of damage caused by

radiation to the device or material. As technology improves, device features get smaller

and use lower energy to operate. This can cause sensitivity to various impinging particles,

which can damage the functionality of the device. The key to reducing the radiation

sensitivity of the device may rely on finding an oxide layer that can remain virtually

unaffected while protecting the underlying semiconductor material and interface.

In addition to permanent degradation of the device, an electrical device can also

experience a number of single event effects (SEE) due to total dose ionization of the

material or device. Soft errors are single event upsets (SEU). Events of this type can be

corrected by reprogramming or resetting the device. This is usually apparent in memory

circuits. It can also result in performance degradation if the error rate is too high. Hard

errors, which are not correctible, include single event latchup (SEL), burnout (SEB), and

gate rupture (SEGR). If a hard error occurs a circuit element can be physically damaged.

2.7 Previous Radiation Studies

In recent years, several studies have been done on radiation of various GaN

devices. Radiation types have included gamma rays, electrons and low-energy protons.

Gamma irradiation at total fluences of 600 Mrads was studied by Luo et al. [49].

At this high dosage the AlGaN/GaN HEMT showed a 45% change in the









transconductance, gm, for varied gate lengths and widths. At a lower dose of 300 Mrad,

minimal changes in the electrical characteristics were observed.

Look et al. [50] conducted an electron radiation study. Energies ranging from 0.7

to 1 MeV were used to induce removal of nitrogen or gallium atoms. Fluences ranging

from 1 to 7 x1016 cm-2 were noted. They concluded that nitrogen Frenkel pair formation

was occurring, resulting in the formation of shallow donors and shallow or deep

acceptors at the same rate. Look's group also concluded that annealing produced Frenkel

pair recombination. This work also supports the donor nature of N vacancies in the GaN.

The effect of proton radiation on AlGaN/InGaN/GaN LEDs was reported by

Osinski et al. [51]. This group used a beam energy of 2 MeV with a total fluence of 1.68

x 1012 protons/cm2. They found a 40% reduction in the output power of the LED even

though the I-V curves of the devices showed very little change. The optical properties of

the irradiated area almost returned to normal when the sample temperature was lowered

to -15 K. Their conclusion was that the proton dosage did not degrade the single

quantum well LEDs and the optical properties remained practically unchanged.

Another group, Khanna et al. [52], studied photoluminescence of proton irradiated

GaN to determine its radiation resistance. Proton beams with an energy of 2 MeV and

varying fluences were used. As expected, they found that GaN was much more radiation

resistant than GaAs. This was determined from PL intensity levels. A reduction in the

intensity of the dominant peak in the PL spectrum of irradiated GaN was noted, implying

that midgap states were created. They attributed the intensity loss as a result of trapped

carriers at radiation-induced defect sites.









In a study of annealing behavior of proton radiated AlGaN/GaN HEMTs, they

found remarkable recovery of the HEMT DC performance. A fluence of lx1014 cm-2 at

an energy of 1.8 MeV was used. After radiation of the HEMT the saturation of the I-V

curves was reduced from 260 mA/mm to about 100 mA/mm. As with other studies the

reduction was attributed to radiation induced traps resulting in removal of free carriers.

Transconductance of the device was also reduced from 80 mS/mm to 26 mS/mm.

However the study showed a gradual increase of electrical characteristics with increasing

annealing temperatures. At a temperature of 800C, the saturation and transconductance

returned to 220 mA/mm and 56 mS/mm respectively [53].

An experiment conducted by Emtsev et al., studied radiation induced defects of n-

type GaN and InN. 1.0 to 1.5 |tm thick layers of hexagonal n-GaN and InN were grown

by MOCVD and plasma-assisted MBE techniques. After pre-radiation characterization

by XRD and Raman spectroscopy, the samples were radiated with protons of 150keV and

then annealed at 500C or 100C for 20 minutes in nitrogen. For the InN, the group found

that the increase in electron concentration was most likely due to the production of

radiation-induced defects with the shallow donor states. Since the production rate did not

change over a wide range of proton dosage, le1015 to le1016cm-2, the defects are believed

to be native defects that can be attributed to immobile nitrogen vacancies in the InN.

Annealing temperatures up to 100C show no change in electron concentration or

electron mobility but did show a pronounced decrease of 30% in electron concentration

between temperatures of 250-300C. At elevated temperatures the annealing behavior

becomes complicated in the sense that a reverse annealing stage of electron concentration

and mobility takes place at approximately 400C. At 500C little change in either









concentration or mobility was noted but the parameters were higher than the non-

irradiated layers of InN. N-GaN: Si showed a substantial decrease in concentration of

charge carriers. It was found that the electron removal rate was dependent on doping

level of the GaN. After annealing to 200C, there is a noticeable decrease in electron

mobility but no significant changes in electron concentration. At temperatures of 300-

400C the mobility continues to drop but at 6000C the electron concentration and

mobility recover substantially. It is noticed that the mobility of the charge carriers

becomes even larger than the initial measured mobility of the n-GaN. Based on this study

the group concluded that in InN, nitrogen vacancies were the likely cause of the increase

of free electrons in the irradiated InN. For n-GaN they concluded that the production rate

of native defects appears to be Fermi-level dependant and that two main recovery stages

of electron concentration were found at intervals of 300-400C and 500-600C [54].

Gaubas et al., studied the radiation effects of semi-insulating layers of GaN grown

on bulk n-GaN/sapphire substrates. The samples were irradiated with 10keV X-ray dose

of 600Mrads and 100keV neutrons with fluences of 5el014 and 1016 cm-2 respectively. A

set of wafers and diode structures were irradiated then characterized post radiation using

Photoluminescence spectroscopy (PL), non-invasive microwave absorption (MVA) and

contact photoconductivity (CPC). The MVA method is based on a pump-probe technique

with optical excitation and microwave absorption by the free carriers, the CPC relies on

measuring the photocurrent decays. It was found that the radiation induces an increase in

the non-radiative trap density, which resulted in a significant decrease in the PL intensity

of the blue, yellow and UV bands. The effects of the disorder caused by the radiation






27


manifested themselves in the long tail CPC measurements as well as MWA decays with a

time stretching factor of .07 [55].









Table 2-1 Properties of Dielectrics Previously Studied for Use with GaN

Material Bandgap Dielectric (s) Melting Point (K)
(eV)


A1N 6.2 8.5 3273

A1203 5.75 12 2319

Ga203 4.4 10 2013

Gd203 8.5 11.4 2668

Ga203(Gd203) 4.7 14.2 2023

SiOx 8-9 3.9 1993

Si3N4 5.0 7.5 2173

MgO 7.3 9.8 3073

Sc203 6.3 11.4 2678




























Figure 2-1 Typical MOSFET
















VG= 0

Gate
Oxide

Source Drain


A
n-type
regions

p-type Substrate




VG< 0


Gate
Oxide

Source A" Drain
B


n-type
regions


p-type Substrate


Figure 2-2 Cross-section illustration of a enhancement mode MOSFET. A) At VG=O the
device is "OFF" B) At VG<0 the device is "ON". Notice the conduction
channel of electrons created with the application of positive gate voltage.






31













Dielectric

SemiconductoI Substrate



Figure 2-3 Cross section of a MOS capacitor.





















+

Figure 2-4 Dipole formation in the presence of an electric field.



















ECR PLASMA
SOURCE
,* I


RHEED Gun


Buffer chamber


Figure 2-5 Sketch of Riber MBE used for oxide growth (after Gila)


SOLID
SOURCES









































A






















0.2

0.4

0.6

0.8
B u


Figure 2-6 AFM images of as received GaN.
received MBE GaN from SVT.


X 0.200 pM/div
2 50.000 nn/div


A) MOCVD GaN from Epitronics B) as

























Figure 2-7 Examples of GaN surfaces before growth. A) The UV-ozone treated surface
of GaN. B) Buffered oxide etched surface of GaN.


A B


Figure 2-8 Photos of RHEED indicating a (1x3) pattern. A) <11-20> crystal direction.
B) <1-100> crystal direction.














CHAPTER 3
EXPERIMENTAL APPROACH

3.1 Alternative Dielectrics

Because of the various limitations of the previously studied dielectrics, alternative

dielectrics are being actively sought. Scandium oxide, magnesium oxide and the ternary

combination of magnesium calcium oxide show promise as candidate dielectrics. MgO

alone as a gate insulator has proven to have a low Dit, but is thermally and structurally

unstable in normal atmosphere. The possible solution to this problem would be to cap the

MgO with a more stable oxide layer or combine the magnesium with another dielectric

material add stability to the oxide. Hence, the purpose of using a magnesium ternary

oxide capped with thermally and structurally stable scandium oxide as dielectric.

The reasons for selecting these particular oxides are as follows:

Lower lattice mismatch to the GaN semiconductor

Wide bandgap

Thermal stability

High dielectric constant

Desirable band alignment with GaN

Each of the proposed dielectrics have bandgaps that are substantially larger than the

bandgap of GaN. In addition to withstanding high operating temperatures, the dielectrics

should also be able to withstand irradiation of protons that are found in caustic

environments of space and low earth orbits.









3.2 Oxide Growth Parameters

3.2.1 Scandium Oxide Growth

The Sc203 samples were grown using the Riber 2300 MBE equipped with a

RHEED system. Scandium oxide exists in a bixbyite structure and was grown using a

standard effusion cell filled with Sc operating at temperatures from 1130C to 11700C

[56,57]. Oxygen was supplied from a RF plasma source which, was kept at 200 watts

forward power. The chamber pressures ranged from 1x10-4 Torr to 5x10-4 Torr. The

substrate temperature was varied from 1000C to 6000C. Higher growth temperatures

produce larger grain sizes. Electron microscopy shows a smooth interface between the

Sc203 and the GaN substrate.

3.2.2 Magnesium Oxide Growth

The magnesium oxide samples were grown from 99.99% pure magnesium and

Knudsen cell temperatures ranging from 3500C to 4000C. Substrate temperatures

between 1000C and 3400C were used. The oxygen was provided by an Oxford RF

plasma source at 300 W forward power. Oxygen pressure was varied from 8x10-6 up to

1x10-5 Torr. In all cases, the sample rotation was kept at 15 rpm.

3.2.3 Magnesium Calcium Oxide Growth

The magnesium calcium oxide samples were grown from the same Mg sources as

that used for the MgO with the addition of 99.999% pure calcium, Ca, from another

Knudsen cell with temperatures ranging from 4500C to 5000C. Substrate growth

temperatures were between 1000C and 3000C. Two growth methods were used: 1)

continuous where all shutters open at once and exposed to the substrate and 2) digital

alloying where alternating layers of MgO and CaO are deposited. Changing the flux of









the metal sources during a continuous growth or changing the timing of the shutter

sequences during digital growth varied the composition of the film. Oxygen pressure was

held at 8x10-6 Torr and the RF plasma source. As in the MgO growth, the sample rotation

was kept at 15 rpm.

3.3 Materials Characterization

3.3.1 Auger Electron Spectroscopy

Auger Electron Spectroscopy (AES) uses a focused electron beam to create

secondary electrons near the surface of a solid sample [58]. Auger electrons are able to

characterize the elemental composition and at times, the chemistry of the surface of

samples. When combined with ion sputtering to gradually remove the surface, AES can

similarly characterize the sample in depth, allowing microanalysis of three-dimensional

regions of the solid samples. Auger is normally non-destructive except during depth

profiling and for materials which are sensitive to the e-beam. The main use of AES is to

discover the elemental composition of inorganic materials or interface compositions.

(Figure 3-1)

3.3.2 Atomic Force Microscopy

Atomic Force Microscopy (AFM) is a real space imaging technique that can

produce topographic images of a surface with atomic resolution in all three dimensions

[58]. Atomic Force Microscopy is a very powerful imaging system since it can study

insulators, semiconductors, conductors, and transparent as well as opaque materials.

Surfaces can also be studied in liquid or in ultra high vacuum and the system uses a sharp

tip mounted on a flexible cantilever. When the tip comes within a few angstroms of the

sample's surface, the repulsive van der Waals forces between the atoms on the tip and









those on the sample cause the cantilever to deflect. Only with an unusually sharp tip and

a flat sample is the lateral resolution truly atomic. (Figure 3-2)

3.3.3 Transmission Electron Microscopy

One of the most powerful microstructural analysis techniques available is

transmission electron microscopy (TEM). From TEM, not only can the microstructure of

an epitaxial film be determined, but also detailed analysis of defects in the film, atomic

imaging of the interface, and an accurate calculation of lattice spacing is determined.

TEM uses a beam of electrons that pass through and interact with a very thin sample to

form an image on the other side of the sample (Figure 3-3) [58]. The interactions

between the atoms in the sample and the electrons produce the contrast seen in the image.

One of the major drawbacks of TEM is the sample preparation required to obtain the

images. The sample must be cut, polished, and thinned to electron transparency (-100

nm) via hand polishing and ion beam milling or by using a focused ion beam (FIB)

system. This is especially difficult for the nitride materials due to their hardness. The

FIB used to make these sample was a FEI Strata DB (Dual Beam) 235 FIB. A JOEL

200CX operating at 200 keV was used for film analysis and JOEL 2010FX operating at

400 keV was used for high-resolution analysis of the interface.

3.3.4 Scanning Electron Microscopy

Scanning Electron Microscopy (SEM) is another technique used to give detailed

material characterization. It uses a rastered electron microscope to highly magnify the

image of the surface of the material [58]. The SEM works by scanning a fine probe of

electrons over the surface of the specimen under vacuum conditions. As electrons

penetrate and react with the surface, an emission of electrons or photons from the surface

is given off. An appropriate detector can then collect the emission and the output can be









used to modulate the brightness of a cathode ray tube (CRT), whose inputs are in sync

with the voltages rastering the electron beam. The image is produced on the CRT and

every point the beam strikes on the sample is mapped directly onto a corresponding point

on the screen. (Figure 3-4)

3.3.5 X-Ray Reflectivity

X-ray Reflectivity (XRR) is a non-destructive technique used to study the structure

and the organization of materials, which are grown as thin films. The range of XRR

includes the submicronic and atomic scales [59]. It can be used to determine the electron

density profile (EDP) and the roughness of interfaces. By measuring the diffuse x-ray

scattering, it can show the interfacial roughness between successive layers. The typical

wavelength of the x-rays are 0.1 nm, which is a very high frequency. As a consequence,

the x-ray interaction with matter can be described by an index of refraction, which

characterizes the change of direction of the x-ray beam when passing from air to a

material. The reflected intensities will be confined in a direction symmetrical from the

incident x-rays beams. This technique is also very sensitive to any defects of flatness,

which makes rough surfaces undesirable when performing this technique. The main

advantages of XRR are the abilities to determine the surface and interface roughness, the

layer thickness, EDP and the structural arrangements of complex films.

3.3.6 Photoluminescence

Photoluminescence (PL) measures physical and chemical properties of materials by

using photons to induce excited electronic states in the material and analyzing the optical

emission as the states relax [58]. The spectral distribution and time dependence of the

emission are related to electronic transition probabilities within the sample, and can be

used to provide qualitative and sometime quantitative information about chemical









composition, structure (i.e. interfaces, bonding, disorder), impurities, and energy transfer.

This technique will allow useful observation of radiation effects to the interface between

the dielectric and semiconductor. Figure 3-5 shows a setup of the PL system that is used

in this study. A helium-cadmium laser is used as the excitation source and scanning

parameters were from 340nm to 800nm.

3.3.7 Hall Effect

The Hall Effect is used to determine carrier concentration of the GaN substrate

before and after thermal anneals and proton irradiation. The hall effect is a result of a

magnetic field being applied perpendicular to the direction of the motion of charged

particles. The magnetic field exerts a force on the current and causes the charged carriers

in the current to split and gather at polar opposite sides of the sample. This enables the

calculation of majority and minority charge carriers. Any change in the concentration in

the carrier numbers can also be observed. The Hall voltage VH is represented by

VH = (RH Ix B)/ d (3-1)

Where RH is the Hall coefficient, Ix is the current in the x-direction, Bz is the magnetic

field in the positive z-direction and d is the thickness of the sample [4].

3.3.8 Current Voltage Analysis

A Hewlett Packard semiconductor parameter analyzer, HP4145B, was used to take

I-V measurements of the MOS devices. The upper and lower current limits were set and

swept from negative to positive. The range of the voltage was increased incrementally

until forward and reverse breakdown were observed. Figure 3-6 depicts an ideal I-V

curve for MOS devices. Note the sharp upward slope of the curve. The reasoning for this

is "turn on" of the device. Once enough voltage is applied to the gate, current is

conducted across the dielectric. The breakdown field strength is given by Equation 3-2.









EBD = V/t (3-2)

Where t is the thickness of the oxide, V is the voltage at the compliance current,

normally 1 mA/cm2. A low breakdown field of the oxide is undesirable in this study,

since it indicates weakness of the dielectric material.


3.1.9 Capacitance Voltage Analysis

Capacitance-Voltage measurement is a critical analysis of electrical devices. C-V

gives information about the fixed oxide charge, interface state density, border trap density

and mobile ion density through graph plots or mathematical computations. All C-V data

was obtained using an automated HP4284A LCR connected to a LabVIEWTM base PC.

The LCR meter supplies a voltage signal of superimposed analog current (AC) and direct

current (DC). Ideal C-V curves for n-type MOS capacitors are shown in figure 3-7. The

C-V curves are frequency- independent in accumulation and depletion but at the on set of

inversion the curves become strongly frequency-dependent. Minimizing the interface

state density, Dit, is also an important aspect of dielectric materials. A high interface state

density of an oxide negates its usefulness as an insulating material for the semiconductor.

3.2 Diode Fabrication

Once the oxides were epitaxially deposited by MBE, the structures were fabricated into

MOS diodes. Scandium oxide was etched using either an Inductively Coupled Plasma

(ICP) or using a hot wet etch of H2SO4 for 6mins. Magnesium Oxide was etched using a

2% solution of phosphoric acid and DI H20 for 20 seconds. The Magnesium Calcuim

Oxide was etched using the same 2% Phosphoric Acid/DI H20 solution. Ohmic contacts

consisted of 200A Ti/700A A1/400A Pt/1000A Au and were deposited by e-beam

evaporation. Gate contacts were Pt/Au, 200A and 1000A respectively. No post anneal









was performed after deposition of the contacts. The finished MOS capacitors are shown

in figure 3-8.

3.3 Proton Radiation Setup and Facility

For proton radiation the samples were taken to the Texas A&M Cyclotron. The oxide and

device samples were radiated at doses of 5x109 cm 2 at energies of 10MeV and 40MeV

under a vacuum in the 10-4 Torr range. The chamber takes at least six minutes to pump

down before radiation can start. The samples were affixed to glass slides and then placed

on the target mounting in the testing chamber as shown in figure 3-9.

The target mount frame can be adjusted in x, y, and z directions and rotated by

computer control. The size of the exposed area is controlled by a shutter like aperture that

can be adjusted horizontally and vertically to insure beam isolation to the desired target

area. Beam uniformity and dosimetry are monitored by an array of five plastic

scintillators located upstream from the target chamber. Beam uniformities of 95% or

better can be achieved. Outside of the testing chamber, an electronic diagnostic setup is

available for data acquisition, in-situ or ex-situ testing (depending on the device) and

quick target changes (Figure 3-10).









Table 3-1 Proposed Oxide Properties.
Bandgap Dielectic Melting Point Mistmatch to
(eV) Constant (s) (K) GaN

Sc203 6.3 11.4 2678 9.2%

MgO 7.3 9.8 3073 -6.5%

MgCaOa 7.4 10.8 2370

GaN 3.4 9.5 2773

a. composition of approx 50/50





























] ELECTRON
MULTIPLIER




IETICSHIELD


GUN


Figure 3-1 Auger Electron Spectroscopy set-up.










A way of sensing the
vertical position of the tip


A course positioning system to
bring to tip into the general
vicinity of the sample.


A feedback
system to
control the
vertical position
of the tip


A piezoelectric scanner
which moves the tip over
the sample (or the sample
under the tip) in a raster
pattern


A sharp tip


_____-V SSample


A computer system that
drives the scanner,
measures data and
converts the data into an
image.


Figure 3-2 Atomic force microscope (after K.K. Harris 2000).













Electron Source


Condenser Lens




Specimen



Obj ective Lens









Back Focal Plane


First Intermediate Image


Intermediate Lens



Second Intermediate Image


Projector Lens







Image

Figure 3-3 TEM setup used to image atomic layers at the film/substrate interface.






















Electron
source


DEFLECTION








Detector giving
modulating signal


./ Secondary Cathode ray
electron signal tube





$S.uinrll)l Screen






Figure 3-4 SEM operation. Electron beam is rastered over the sample producing
secondary electrons (after S.M. Donovan 1999).






48






Monochrometer


Figure 3-5 Photoluminescence setup



























Figure 3-6 Ideal MOS I-V curve


2.4


2.2


2.0


1.8


1.6


1.4


1.2 L
-1.5


-1.0 -0.5 0.0 0.5


Voltage (arb. units)


Figure 3-7 Ideal C-V curves of n-type MOS (after Johnson)







+


+


+


ED


Figure 3-8 Finished MOS capacitors design.


mmmE 3
S3 S 3
wwwww


SSSQ
SSSSS
SSSSS


E E EE
S3 S3 SS
S3 SSQ


EE EEl E
S3~1; S3 S3S3S






































l Beam Shutters










B

Figure 3-9 Test Chamber for Radiation. A) Mounted samples in test chamber. B)
Shutters of beam.










Beam
Front s Viewer
Shutter


Rear Shutter
and


Target Camera .



'- .
Target
Chamber

In-Air
PosMtoM7
6<.*
'Y ']


* Degradea


C-- -


Control
Interface


/
Q


Experimental
Area


Figure 3-10 Radiation Test Facility at Texas A&M University.














CHAPTER 4
RADIATION AND THERMAL STABILITY OF VARIOUS TYPES OF GALLIUM
NITRIDE

Various types of GaN samples were tested pre- and post-proton radiation using

photoluminescence (PL). The effects of rapid thermal annealing (RTA) were also

investigated. The types of gallium nitride that were tested included n-type GaN, a high

resistivity GaN (known as u-GaN), MBE grown GaN, and p-type GaN. PL was used to

determine the initial luminescence state of the GaN and monitor changes in the

luminescence profile after proton radiation and thermal annealing. RTA was used to

determine if the radiation damage to the sample could be corrected by heating the sample

to various temperatures.

4.1 Characterization of MOCVD N-type GaN

As grown commercial MOCVD GaN, supplied by Uniroyal Optoelectronics

(UOE), was probed using PL. Figure 4-1 shows the initial GaN luminescence profile at

room temperature (300K) and low temperature (15K). At 362nm(3.42eV), the primary

transition or bandedge of the GaN is noted in addition to a defect band in the 450nm to

700nm range, which is related to various midgap defects in the gallium nitride. This

yellow defect band gives n-GaN its distinctive luminescence color. Although the origin

of this yellow band is heavily debated, the defect origins are most likely grown-in

nitrogen vacancies. Notice at 15K the large defect emission from 450nm to 700nm has

been suppressed. According to Khanna et al., the suppression of the defect band at low









temperatures is a result of reduced non-radiative transitions and a lack of available

phonons [52].

The UOE n-GaN samples were irradiated at two different energies of 10MeV and

40MeV with a total particle dose of 5 xl09cm-2, which equals approximately 10 years of

mission time in space applications. Figures 4-2 and 4-3 show the post radiation damage

to the GaN in comparison to the initial scan at 300K and 15K respectively. At 300K a

significant decrease in luminescence intensity is noticed as well as a slight shift in the

bandedge peak to 354nm. Post radiation at 15K shows an increase in the defect band of

the profile which can be attributed to the increased amount of defects within the bulk

material that have now be come radiative centers for electron -hole recombination. The

bandedge is significantly decreased also due to the introduction of non-radiative defects.

Also due to the radiation damage the UOE n-GaN samples luminescenced a pale green

color and not the customary yellow that characterizes GaN. Based on previous work done

by other groups, the reason for this decrease in intensity and change in color is an

increase in lattice defects.

After a year, the samples were rescanned with PL to determine the luminescence

profile. Figure 4-4 shows a comparison to the initial GaN scans and the post radiated

GaN. After one year, the defect luminescence returns to its pre-irradiated state.

Similarly, there is a partial though not complete recovery of the bandedge emission.

This change in the profile from the immediate post radiation scans implies either long

life-time traps have relaxed or surface carrier loss has been suppressed via oxidation or

adsorption of adventitious species such as carbon. The samples still luminescence with a









pale green-yellow and the PL profile was not completely returned to the original scan

which shows that not all the changes caused by irradiation were corrected.

In order to determine if the radiated samples could be returned to their initial profile

by thermal treatment, the UOE n-GaN samples were annealed using an RTA setup with

varying temperatures ranging from 200C to 9000C and held for 60 seconds at each

temperature. After each temperature anneal, the GaN samples were rescanned using PL.

In order to better monitor the changes in the n-GaN induced by annealing, a non-

irradiated n-GaN sample from the same wafer as the irradiated samples was also annealed

and PL scanned. Figures 4-5 through 4-7 shows a comparison of the annealed radiated

and non-radiated samples of n-GaN. After annealing at 2000C a significant increase in PL

intensity was observed for all samples but at 300C the intensity decreases well below the

original level. It seems unlikely that diffusion of nitrogen and gallium atoms to vacancies

or interstitial sites within the GaN lattice has occurred at such a low temperature. Thus, it

is possible that some surface phenomena are responsible, though the exact mechanism is

as yet unknown. At higher temperatures of 400C to 900C, no large changes to the PL

profiles in either intensity or bandedge shifts were observed. Based on figure 4-8, the

GaN has reached a fairly steady configuration of defects within the lattice and that no

more recovery or improvement of the samples can be achieved without completely

damaging the GaN substrate. Figure 4-8 is a calculated trend of the annealed sample

changes based on ratio comparisons of normalized bandedge peaks of all annealed GaN

to the GaN control peak intensity of 1.4812. After 600C, there is a gradual decrease in

bandedge intensity indicating that further annealing is damaging the optical properties of

the GaN.









After reviewing the results of the first UOE n-GaN, a further study of other types of

GaN with different growth recipes and growth systems was undertaken in order to further

investigate the phenomena at 200C. Another UOE n-GaN sample from another wafer,

which will be referred to as the 2nd UOE n-GaN sample, was scanned with PL, radiated at

10MeV and 40MeV and annealed from 200C to 9000C to determine the 2nd UOE n-GaN

trend. Figure 4-9 shows the pre and post radiation PL profiles of the 2nd UOE n-GaN. In

comparison, both UOE n-GaN samples show the same basic trends in the radiation and

thermal experiments. Figures 4-10 through 4-12 show the radiation and thermal effects

on PL profiles for the 2nd UOE n-GaN sample. All of the data from the 2nd UOE n-GaN

sample was collected immediately after radiation. As noted before, after annealing at

200C a large increase in the PL intensity is observed. Thus far, this phenomenon has

shown to be unique to all UOE GaN (n-GaN and u-GaN) samples and was not observed

for any of the other GaN types, which will be discussed in later sections. This

phenomenon could have various reasons, such as sample preparation, growth parameters

for UOE GaN, surface effects or material doping, which are unknown and proprietary

information at the present time. In figure 4-13, the normalized bandedge peak intensity

versus annealing temperature is shown for the 2nd UOE n-GaN sample. The trend in the

plot generally follows the first UOE n-GaN with no significant changes in the profiles

after 400C. For both samples, annealing was not effective in completely restoring the

spectra to the pre-irradiated profile, and after annealing at 9000C the luminescence ratio is

almost equal to that of the radiated sample ratio. At the higher temperatures, it is likely

that nitrogen is being driven out of the sample and leaving behind nitrogen defects that

essentially creates non- radiative centers.









In addition, n-GaN grown by an MOCVD system at the University of Florida (UF)

was also studied and underwent the same experiments as the UOE n-GaN to explore the

potential variability in MOCVD GaN. The GaN grown at UF sample will be referred to

as UF n-GaN through out the paper. Figure 4-14 shows the pre and post radiation PL

scans, like the UOE n-GaN, irradiation of the UF n-GaN produced a significant drop in

PL intensity. Figures 4-15 through 4-17 shows the results of rapid thermal annealing from

200C to 900C for each UF n-GaN sample. Figure 4-18 plots the bandedge peaks

normalized to 25.069 in intensity. Unlike the UOE n-GaN, the non-irradiated UF n-GaN

shows a significant drop in both bandedge and defect luminescence for all annealing

temperatures. Annealing of the irradiated material produces little to no improvement in

the intensity of the PL.

4.2 Characterization of MOCVD U-type GaN

As with the n-GaN samples, a set of UOE verses UF u-GaN was tested and

compared. Generally u-GaN has lower luminescence intensity due to a reduction in free

carriers. Figure 4-19 shows the changes in the PL scan after proton radiation at 10MeV

and 40MeV. Notice that there is a severe reduction in the already low peak intensity. The

broadness of the bandedge peak indicates the presence of defects in the gap of the GaN.

Figures 4-20 through 4-22 depicts the effects of thermal annealing of the samples. In each

of the samples there is an increase at 200C, which, as mentioned before in previous

section, is mostly likely surface effects of the GaN since the intensity continuously drops

with increasing annealing temperatures beyond the temperature. Normalizing the

bandedges peak intensity to the control sample bandedge shows the trend of increasing

and decreasing intensity with annealing temperature (figure 4-23). At 400C there is a

severe decrease in intensity and no additional drastic change or recovery of the samples.









The UF u-GaN shows higher peak intensity than the UOE u-GaN possibly

indicating a better quality sample. Figure 4-24 shows the pre and post radiation scans of

the UF u-GaN and in figures 4-25 through 4-27 the thermal annealing is shown. The main

trend of both the UOE u-GaN samples and the UF u-GaN is that at 4000C the control

samples and the radiated samples are about the same in peak intensity and continue in a

linear fashion to 9000C. Figure 4-28 exhibits this trend in the UF u-GaN. Radiation

reduced the bandedge but only slightly reduced the defect luminescence. As with the UF

n-GaN, annealing at all temperatures reduce the PL intensity.

4.3 Characterization of MBE Grown Ga-Polar GaN

MBE gallium nitride is known to be highly resistive mostly likely due to self-

compensation of the Ga and exhibit for weak PL emissions due to lack of free carriers

available in the samples for emission transitions. Figure 4-29 shows the comparison of

the pre and post radiation scans of the Ga-polar GaN sample. The bandedge peak after

irradiation is no longer visible in the PL scan mostly indicating a large increase in

optically active defects in the samples or degradation of the surface. After annealing each

sample (figures 4-30 through 4-32), there was a slight increase in the peak intensity at the

higher temperatures but it never recovers to the original peak intensity. Figure 4-33

shows the normalized bandedge peak intensities to the control versus the annealing

temperatures.

4.4 Characterization of MOCVD P-type GaN

The comparison ofUF MOCVD grown p-type GaN was studied with PL,

annealing, and Hall to determine the behavior of the sample after proton irradiation and

thermal testing. The samples were taken from the same wafer but one half of the wafer

received a post activation anneal of the Mg implantation at 7500C for 30 seconds in the









RTA. The as grown p-GaN will be referred to as P-GaN and the part of the wafer that

received the post growth activation anneal will be referred to as Activated P-GaN. In

figure 4-34 is the PL comparison of each control sample. Figures 4-35 and 4-36 shows

the pre and post PL scans of the p-GaN and activated p-GaN samples. Note the difference

in peak intensity of the control samples and then the observed decrease in intensity of the

radiated p-GaN samples but the increase in the activated p-GaN samples. It is possible

that the radiation in the activated samples has caused ionization of the Mg acceptors.

Figures 4-37 through 4-38 show the PL scans after annealing each of the p-GaN (un-

activated and activated) control samples and figures 4-39 and 4-42 captures the trend of

the 10MeV and 40MeV radiated p-GaN (un-activated and activated) samples. Each of the

samples experienced a sharp decline in intensity at 4000C and again at 600C. They each

increased in peak intensity at 9000C which mostly likely can be attributed to defect

annealing. Each of the un-activated p-GaN samples generally follows the same trend with

or without radiation, whereas the activated p-GaN shows a larger spread in the ratios of

the normalized bandedge peaks (figures 4-43 and 4-44). Figure 4-45 compares the two

plots of bandedge peaks versus annealing temperature where it exhibits the difference in

the changes of the samples. The as grown p-GaN look much more linear than the

activated p-GaN.

4.5 Summary

Based on the data collected, some general trends for the GaN can be noted. First is

that radiation of the samples decrease the PL luminescence intensity. Second, an increase

of PL intensity at 2000C has shown to be unique to the UOE GaNs and as-grown P-GaN.

After approximately 400C, the drastic decline in peak intensity stopped and became less






60


significant in change with higher temperatures. These defects become non-radiative

centers and trap carriers that contribute to optical transitions in the bandgap of the GaN.










- 1st UOE n-GaN Control @ 300K
-- 1st UOE n-GaN Control @ 15K


400 500 600 700


800


Wavelength

Figure 4-1 Initial PL scans of non-radiated 1st UOE n-GaN at temperatures 300K and
15K.


0.8-

0.7-

0.6-

0.5-

0.4-

0.3-

0.2-


0.1

0.0-

300






62



0.7 -- 1st UOE n-GaN Control

---Rad 10MeV
0.6- Rad 40MeV

0.5

'" 0.4
c

0.3 1
(J
0.2 -

0.1 ,

0.0

300 400 500 600 700 800
Wavelength


Figure 4-2 Room Temperature PL analysis of 1st UOE n-GaN before and after radiation.






63









1st UOE n-GaN Control
Rad 10MeV
Rad 40MeV












i i- -


' I I
500 600

Wavelength


700


8 I
800


Figure 4-3 15K PL analysis of 1st UOE n-GaN before and after radiation.


3.5-


3.0-


2.5-


2.0-


1.5-


1.0-


0.5-


0.0

300


400
400












-- 1st UOE n-GaN Control
-- Radiated 10MeV
Radiated 40MeV
-- 10MeV 1 year later
40MeV 1 year later


400 500 6I0 70
400 500 600 700


800


Wavelength



Figure 4-4 Comparison of 300K 1 st UOE n-GaN PL scans, pre-radiation, immediately
post radiation, and post radiation 1 year later.


0.8 -


0.7-


0.6-


0.5-


0.4-


0.3-


0.2-


0.1-


0.0 -


300


_











-- 1st UOE n-GaN Control
--Annealed 200C
Annealed 300C
-- Annealed 400C
Annealed 600C
-- Annealed 800C
Annealed 900C


1.5





1.0





0.5





0.0



300


800
800


Wavelength


-- 1st UOE n-GaN Control
--Annealed 200C
Annealed 300C
--Annealed 400C
Annealed 600C
-- Annealed 800C
Annealed 900C


350 360 370 380
Wavelength
B

Figure 4-5 PL analysis of annealed un-radiated 1st UOE n-GaN. A) PL Spectra B)
Bandedge inset.


700
700


400
400


I
500


600
600


(n


0- 05




00













-- 1st UOE n-GaN Control
--Radiated 10MeV
Annealed 200C
-- Annealed 300C
Annealed 400C
- Annealed 600C
Annealed 800C
-- Annealed 900C


600
600


700


800
800


Wavelength


- 1st UOE n-GaN Control
-- Radiated 10MeV
Annealed 200C
--Annealed 300C
Annealed 400C
--Annealed 600C
Annealed 800C
- Annealed 900C


Figure 4-6 PL analysis of annealed 10MeV radiated 1st UOE n-GaN A) PL Spectra B)
Bandedge inset (below).


2.1-


1.4-





0.7-


0.0-


400
400


500
500


>,
0)
rO

oa 07




00


Wavelength











-- 1st UOE n-GaN Control
-- Radiated 40MeV
Annealed 200C
--Annealed 300C
Annealed 400C
--Annealed 600C
Annealed 800C
--Annealed 900C


700
700


800
800


Wavelength


- 1st UOE n-GaN Control
- Radiated 40MeV
Annealed 200C
--Annealed 300C
Annealed 400C
-- Annealed 600C
Annealed 800C
- Annealed 900C


Wavelength

B

Figure 4-7 PL analysis of annealed 40MeV radiated 1st UOE n-GaN. A) PL Spectra B)
Bandedge inset.


1.0-





0.5-


0.0-


300


400
400


500
500


600
600


15




10

(0
t-
as
. 05




00












--1st UOE n-GaN Control
-- Radiated 10MeV
Radiated 40MeV


1.5-
1.4-
1.3-
1.2-
1.1-
1.0-
0.9-
0.8-
0.7-
0.6-
0.5-
0.4-
0.3-
0.2-
0.1-


1000


Temperature


Figure 4-8 Normalized bandedge peaks of all annealed 1st UOE GaN to the 1st UOE GaN
control peak intensity of 1.4812 vs. the annealing temperatures. 0 temperature
indicates the initial state of the GaN.


200









69



12


S10 2nd UOE n-GaN Control
--Radiated 10MeV
Radiated 40MeV
8


( 6


a, 4


2


0-


300 400 500 600 700 800
Wavelength


Figure 4-9 Pre and post radiation PL analysis of 2nd UOE n-GaN.














--2nd UOE n-GaN Control
-- Annealed 200C
Annealed 400C
-- Annealed 600C
Annealed 800C
-- Annealed 900C


700


800
800


Wavelength


- 2nd UOE n-GaN Control
-- Annealed 200C
Annealed 400C
- Annealed 600C
Annealed 800C
-- Annealed 900C


350 360 370 380
Wavelength


Figure 4-10 PL analysis of un-radiated annealed 2nd UOE n-GaN. A) PL Spectra B)
Bandedge inset.


400
400


500
500


600
600











-- 2nd UOE n-GaN Control
-- Radiated 10MeV
Annealed 200C
-- Annealed 300C
Annealed 400C
-- Annealed 600C
Annealed 800C
-- Annealed 900C


I I 7 I
500 600 700


Wavelength


2nd UOE n-GaN Control
Radiated 10MeV
Annealed 200C
Annealed 300C
Annealed 400C
Annealed 600C
Annealed 800C
Annealed 900C


350 360 370 380
Wavelength


Figure 4-11 PL analysis of annealed 10MeV radiated UOE n-GaN. A) PL Spectra B)
Bandedge inset


I
400


8




C4
_.




0













--2nd UOE n-GaN Control
-- Radiated 40MeV
Annealed 200C
-- Annealed 300C
Annealed 400C
-- Annealed 600C
Annealed 800C
-- Annealed 900C


0-


300


Wavelength


2nd UOE n-GaN Control
Radiated 40MeV
Annealed 200C
-Annealed 300C
Annealed 400C
-Annealed 600C
Annealed 800C
-Annealed 900C


350 360 370 380
Wavelength


Figure 4-12 PL analysis of annealed 40MeV radiated UOE n-GaN. A) PL Spectra B)
Bandedge inset















-1- 2nd UOE N-GaN Control
1.2 -
S-*-Rad 10MeV
Rad 40MeV
1.0


c, 0.8
0

( 0.6- \
0)
-c,
-0
S0.4


0.2 -


0.0 -


0 200 400 600 800 1000

Temperature




Figure 4-13 Normalized bandedge peaks of annealed 2nd UOE n-GaN to the 2nd UOE n-
GaN control peak intensity of 12.354 vs. the annealing temperatures.






74




25 -- UF n-GaN Control
-- Radiated 10MeV
Radiated 40MeV
20



S15-
cn
Cl)

_ 10



5



0-


300 400 500 600 700 800

Wavelength


Figure 4-14 Pre and post radiation PL analysis ofUF n-GaN.











-- UF n-GaN Control
-- Annealed 200C
Annealed 400C
-- Annealed 600C
Annealed 800C
-- Annealed 900C


Wavelength


360 370
Wavelength


Figure 4-15 PL analysis of annealed un-radiated UF n-GaN. A) PL Spectra B) Bandedge
inset.


0-

300


18

0)
-c


S9




0











-- UF n-GaN Control
-- Radiated 10MeV
Annealed 200C
-- Annealed 300C
Annealed 400C
-- Annealed 600C
Annealed 800C
-- Annealed 900C


I II I I I
400 500 600 700 800

Wavelength


-UF n-GaN Control
-Radiated 10MeV
Annealed 200C
-Annealed 300C
Annealed 400C
-Annealed 600C
Annealed 800C
-Annealed 900C


350 360 370 380
Wavelength


Figure 4-16 PL analysis of annealed 10MeV radiated UF n-GaN. A) PL Spectra B)
Bandedge inset


0-

300











-- UF n-GaN Control
-- Radiated 40MeV
Annealed 200C
-- Annealed 300C
Annealed 400C
--Annealed 600C
Annealed 800C
-- Annealed 900C


Wavelength


- UF n-GaN Control
- Radiated 40MeV
Annealed 200C
-- Annealed 300C
Annealed 400C
S-- Annealed 600C
Annealed 800C
-- Annealed 900C


340 350


360
Wavelength


370 380


Figure 4-17 PL analysis of annealed 40MeV radiated UF n-GaN. A) PL Spectra B)
Bandedge inset


300

300


400


500






















-- UF N-GaN Control
-*-Rad 10MeV
Rad 40MeV


20 40
200 400


1000


Temperature




Figure 4-18 Normalized bandedge peaks of annealed UF n-GaN to the UF n-GaN control
peak intensity of 25.069 vs. the annealing temperatures.


1.0-



0.8-



0.6-



0.4-



0.2-



0.0-


III













0.12

-- UOE u-GaN Control
0.10 -- Radiated 10MeV

Radiated 40MeV
0.08

c2)
. 0.06
C

co
iD 0.04


0.02


0.00
I I I I I
300 400 500 600 700 800

Wavelength


Figure 4-19 Pre and Post Radiation of UOE u-GaN.













0.12-


0.10-


0.08-


0.06-


0.04-


-- UOE u-GaN Control
-- Annealed 200C
Annealed 400C
-- Annealed 600C
Annealed 800C
-- Annealed 900C


0.02


0.00
I I I I I
300 400 500 600 700 800
Wavelength


Wavelength
B

Figure 4-20 PL analysis of annealed un-radiated UOE u-GaN. A) PL Spectra B)
Bandedge inset.













-- Control UOE u-GaN
-- Radiated 10MeV
Annealed 200C
--Annealed 300C
Annealed 400C
--Annealed 600C
Annealed 800C
--Annealed 900C


Wavelength


350 360 370 380
Wavelength

B

Figure 4-21 PL analysis of annealed 10MeV radiated UOE u-GaN. A) PL Spectra B)
Bandedge inset.


0.12-


0.10-


0.08-


0.06-


0.04-


0.02-


0.00-















-- Control UOE u-GaN
-- Radiated 40MeV
Annealed 200C
-- Annealed 300C
Annealed 400C
-- Annealed 600C
Annealed 800C
-- Annealed 900C


400 500 I
400 500


600
600


700


800
800


Wavelength


-Control UOE u-GaN
- Radiated 40MeV
Annealed 200C
-Annealed 300C
Annealed 400C
-Annealed 600C
Annealed 800C
-Annealed 900C


340 350 360
Wavelength


370 380


Figure 4-22 PL analysis of annealed 40MeV radiated UOE u-GaN. A) PL spectra B)
Bandedge inset.


0.12-



0.10-



0.08-
>,
C
(D 0.06-



a 0.04-



0.02-


0.00-


012-





008-



c

0- 004-




n An
















1.2


1.0 -- UOE U-GaN Control
-*-Rad 10MeV
Rad 40MeV
n 0.8


( 0.6-
0)
(U
S0.4


0.2 -


0.0

I I I I I I
0 200 400 600 800 1000

Temperature

Figure 4-23 Normalized bandedge peaks of annealed UOE u-GaN to the UOE u-GaN
control peak intensity of 0.1162 vs. the annealing temperatures.