Citation
Reliability Study of Gan-Based High Electron Mobility Transistors

Material Information

Title:
Reliability Study of Gan-Based High Electron Mobility Transistors
Creator:
Liu, Lu
Place of Publication:
[Gainesville, Fla.]
Florida
Publisher:
University of Florida
Publication Date:
Language:
english
Physical Description:
1 online resource (197 p.)

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Chemical Engineering
Committee Chair:
REN,FAN
Committee Co-Chair:
PEARTON,STEPHEN J
Committee Members:
TSENG,YIIDER
GILA,BRENT P
Graduation Date:
12/13/2013

Subjects

Subjects / Keywords:
Direct current ( jstor )
Dosage ( jstor )
Drains ( jstor )
Electric current ( jstor )
Electric fields ( jstor )
Electric potential ( jstor )
Electrons ( jstor )
Proton irradiation ( jstor )
Protons ( jstor )
Threshold currents ( jstor )
Chemical Engineering -- Dissertations, Academic -- UF
fet -- gan -- proton-irradiation -- reliability
Genre:
bibliography ( marcgt )
theses ( marcgt )
government publication (state, provincial, terriorial, dependent) ( marcgt )
born-digital ( sobekcm )
Electronic Thesis or Dissertation
Chemical Engineering thesis, Ph.D.

Notes

Abstract:
Compound semiconductors are III-V semiconductor nitrides, such as AlN, GaN and AlGaN, which have attracted plenty of attention due to their extraordinary material properties, such as high mobility, high saturation velocity and power density, and become a promising alternative for microwave power device. They have demonstrated excellent performance in such applications as UV detectors, UV and visible light emitting diodes (LEDs), microwave power amplifications, satellite, radar and wireless communication systems. However, the main impediment of this technology is lack of reliability, and some relative studies and degradation mechanisms have been proposed, such as hot-electron-induced trap degradation, inverse piezoelectric effect and electric-field driven mechanism under off-state conditions. In order to improve the reliability of GaN HEMTs technology, it is essentially important to understand the nature of device degradation. In this dissertation, the effects of source field plates, gate metallization, passivation layers and buffer structures on the dc/rf performance and reliability of AlGaN/GaN high electron mobility transistors (HEMTs) under off-state stress conditions were investigated using step-stress cycling. 19 The source field plate enhanced the drain breakdown voltage from 55V to 155V and the critical voltage for off-state gate stress from 40V to 65V, relative to devices without the field plate. Transmission electron microscopy (TEM) results showed the presence of pits that appeared on both source and drain side of the gate edges, which was attributed to the inverse piezoelectric effect. The significant improvements of dc performance and reliability of AlGaN/GaN high electron mobility transistors (HEMTs) had been achieved by employing Pt/Ti/Au instead of the conventional Ni/Au gate metallization. The critical voltage of electrical step-stress was increased more than 100% for HEMTs with Pt/Ti/Au gate metallization as compared to HEMTs with Ni/Au gate metallization. Another issue related to device reliability is current collapse, which limits the performance of HEMTs in high power and high frequency applications. We had investigated Al To understand the trap behavior in GaN-based HEMTs, a trap-analysis method was developed, and the effect of off-state electrical stress on the trap densities was studied as well. Two traps with different activation energies at temperature range of 300-493K and 493-573K were identified by temperature dependence of sub-threshold slope measurement. In addition, the effects of defect densities on the reliability of AlGaN/GaN HEMTs were also studied. The device with three different types of buffer 20 2O3 and HfO2 deposited with an atomic layer deposition system and conventional, thick (200 nm) plasma enhanced chemically vapor deposited SiNx. The latter is found to be the most effective in reducing drain current loss during gate lag measurements in both single and double pulse mode, but also reduces fT and fMax through additional parasitic capacitance. layers, including an AlGaN/GaN composite layer, or 1 or 2 µm GaN thick layers, were fabricated and their reliability compared. The HEMTs with the thick GaN buffer layer showed the lowest critical voltage (V For space-based applications, electronic components may suffer radiation damage from high fluxes of energetic particles. Hence, the radiation hardness of AlGaN/GaN and InAlN/GaN HEMTs was investigated. The dependence of dc and rf characteristics and reliability of GaN-based HEMTs on proton irradiation energies and doses were examined by dc, small and large signal, pulse measurements, as well as electrical step-stress. GaN-based HEMTs showed a remarkable resistance to high energy proton-induced degradation and improved device reliability. The critical voltage increased from 15V to 45V for HEMTs post to 5MeV proton irradiation at a dose of 5 × 10cri) during off-state drain step-stress, but this was increased by around 50 and 100% for devices with the composite AlGaN/GaN buffer layers or thinner GaN buffers, respectively. 15 cm-2. ( en )
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis:
Thesis (Ph.D.)--University of Florida, 2013.
Local:
Adviser: REN,FAN.
Local:
Co-adviser: PEARTON,STEPHEN J.
Statement of Responsibility:
by Lu Liu.

Record Information

Source Institution:
UFRGP
Rights Management:
Copyright Liu, Lu. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
906819393 ( OCLC )
Classification:
LD1780 2013 ( lcc )

Downloads

This item has the following downloads:


Full Text

PAGE 1

1 RELIABILITY STUD Y OF GAN BASED HIGH ELECTRON MOBILITY TRANSISTORS By LU LIU 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 2013

PAGE 2

2 2013 Lu Liu

PAGE 3

3 D edicated to my loving mother and wife for their love and constant support

PAGE 4

4 ACKNOWLEDGMENTS First of all, I would like to appreciate my advisor and supervisory committee chair, Dr. Fan Ren. Thanks for his broad and profound knowledge, patience and encouragement through my Ph.D. study. The completion of this work would not be possible without the guidance from Dr. Fan Ren. He has always been there t o provide technical and personal support whenever I had problems. He has treated me like his kid and the group like a family, and provided an excellent academic atmosphere for me to perform my research. His education and dedication in the past four years have been priceless and will benefit my future life. I also would like to express my sincere gratitude to the rest of my committee members, Dr. Stephen J. Pearton, Dr. Brent P. Gila and Dr. Yiider Tseng. Thanks for their suggestions and efforts to improve m y proficiency in my research area. The reason I was able to finish this work was the support and assistance of former group members Drs. Chihyang Chang, Nimo Chen, Byung Hwan Chu and ChienFong Lo, as well as current group members Tsung Sheng Kang, Xiaoti e Wang, Yuyin Xi, Yuxi Wang, Ya Hsi Hwang and Camilo Velez in Chemical Engineering; Dr. Erica Douglas, Dr. David Cheney, Patrick Whiting and Ray Holzworth in materials science; Bill Lewis, David Hays and Joel Fridmann at Nanoscale Research Facility at Univ ersity of Florida. Thanks for all your support and caring through good times to bad times in the past few years. Special thanks to Department of Chemical Engineering at the University of Florida giving me this opportunity to pursue my PhD degree in this su nshine state, and my great thanks to department stuff Carolyn Miller in payroll for purchasing, Dennis Vince

PAGE 5

5 and Jim Hinnant in mechanical and electronic shops for tool maintenance and troubleshooting. Finally, I would like to thank my family for their lov e and support. I want to thank my wife, Summer Wang, for always standing beside me. Her love and encouragement allow me to finish this journey and she is the most brilliant rainbow in Florida sky. I hope that this work makes you proud.

PAGE 6

6 TABLE OF CON TENTS page ACKNOWLEDGMENTS .................................................................................................. 4 LIST OF TABLES ............................................................................................................ 9 LIST OF FIGURES ........................................................................................................ 10 LIST OF ABBREVIATIONS ........................................................................................... 15 ABSTRACT ................................................................................................................... 17 CHAPTER 1 INTRODUCTION .................................................................................................... 19 1.1 Introduction to GaN based High Electron Mobility Transistors (HEMTs) .......... 1 9 1.2 Motivation ......................................................................................................... 20 1.3 Background ...................................................................................................... 20 1.3.1 GaN based HEMTs ................................................................................. 20 1.3.2 Origin of 2DEG ........................................................................................ 22 1.3.3 The operation of FET .............................................................................. 22 1.3.4 Substrates ............................................................................................... 23 1.4 GaN Reliability Issues ...................................................................................... 25 1.4.1 Hot electron Effect .................................................................................. 25 1.4.2 Inverse Piezoelectric Effect ..................................................................... 26 1.4.3 Trapping Effect ........................................................................................ 27 1.5 Dissertation Outline .......................................................................................... 28 2 THE EFFECTS OF SOURCE FIELD PLATE ON THE CHARATERISTICS OF OFFSTATE, STEP STRESSED ALGAN/GAN HIGH ELECTRON MOBILITY TRANSISTOR S ...................................................................................................... 35 2.1 Introduction to Field Plate ................................................................................. 35 2.2 Experimental ..................................................................................................... 36 2.2.1 Material Growth ....................................................................................... 36 2.2.2 Material Growth and HEMTs Fabrication ................................................ 36 2.2.3 Electrical Stepstress .............................................................................. 37 2.2.4 Device Characterization and Simulation ................................................. 38 2.3 Results and Discussion .................................................................................... 38 2.4 Summary .......................................................................................................... 41 3 THE EFFECT OF GATE METALLIZATION ON THE RELIABILITY OF ALGAN/GAN HIGH MOBILITY TRANSISTORS ..................................................... 50

PAGE 7

7 3.1 Overview of Gate Metallization of AlGaN/GaN HEM Ts .................................... 50 3.2 Experimental ..................................................................................................... 51 3.2.1 Material Growth and HEMTs Fabrication ................................................ 51 3.2.2 HEMTs Characterization ......................................................................... 51 3.3 Results and Discussion .................................................................................... 52 3.4 Summary .......................................................................................................... 55 4 INVESTIGATING THE EFFECT OF OFFSTATE STRESS ON TRAP DENSITIES IN ALGAN/GAN HIHG MOBILITY TRANSISTORS ............................ 62 4.1 Background ...................................................................................................... 62 4.2 Experimental ..................................................................................................... 63 4.3 Results and Discussion .................................................................................... 63 4.4 Summary .......................................................................................................... 68 5 COMPARISON OF PASSIVATION LAYERS FOR ALGAN/GAN HIGH MOBILITY TRANSISTORS ..................................................................................... 75 5.1 Background ...................................................................................................... 75 5.2 Experimental ..................................................................................................... 76 5.3 Results and Discussion .................................................................................... 77 5.4 Summary .......................................................................................................... 80 6 EFFECT OF BUFFER DESIGN ON THE RELIABILIT Y OF ALGAN/GAN HIGH ELECTRON MOBILITY TRANSISTORS ................................................................ 90 6.1 Background ...................................................................................................... 90 6.2 Experimental ..................................................................................................... 92 6.3 Results and Discussion .................................................................................... 92 6.4 Summary .......................................................................................................... 98 7 EFFECTS OF PROTON IRRADIATION ON GANBASED HIGH ELECTRON MOBILIT Y TRANSISTORS ................................................................................... 109 7.1 The Effects of Proton Irradiation on the Reliability of InAlN/GaN High Electron Mobility Transistors .......................................................................... 109 7.1.1 Background ........................................................................................... 109 7.1.2 Experimental ......................................................................................... 111 7.1.3 Results and Discussion ......................................................................... 112 7.1.4 Summary .............................................................................................. 115 7.2 The Effects of Proton Energy on the Degradation of AlGaN/GaN High Electron Mobility Transistors .......................................................................... 116 7.2.1 Background ........................................................................................... 116 7.2.2 Experimental ......................................................................................... 117 7.2.3 Results and Discussion ......................................................................... 118 7 .2.4 Summary .............................................................................................. 124 7.3 The Effects of Proton Dose on the Degradation of AlGaN/GaN High Electron Mobility Transistors .......................................................................... 125

PAGE 8

8 7.3.1 Background ........................................................................................... 125 7.3.2 Experimental ......................................................................................... 126 7.3.3 Results and Discussion ......................................................................... 127 7.3.4 Summary .............................................................................................. 133 8 LASER ABLATION OF SILICON CARBIDE ......................................................... 161 8.1 Background .................................................................................................... 161 8.2 Experimental ................................................................................................... 163 8.3 Results and Discussion .................................................................................. 164 8.4 Summary ........................................................................................................ 168 9 CONCLUSIONS ................................................................................................... 176 LIST OF REFERENCES ............................................................................................. 182 BIOGRAPHICAL SKETCH .......................................................................................... 197

PAGE 9

9 LIST OF TABLES Tabl e page 1 1 Summary of the properties of Al2O3, SiC, Si, and GaN and the challenges. ....... 34 3 1 Summary of subthreshold sl ope and ON/OFF ratio of AlGaN/GaN HEMTs with Pt/Ti/Au and Ni/Au gate metallization before and after off state electric step stress .......................................................................................................... 61 4 1 Sub threshold drain leakage current, the slope of the sub threshold drain current and drain current on/off ratio of the HEMTs before and after the off state drain voltage stepstress ............................................................................ 74 4 2 Room and higher Temperature drain saturation current, maximum gm, drain current ON/OFF ratio, isolation current at 10V for both reference and stressed samples ................................................................................................ 74 5 1 Summary of dc characteristics of unpassivated and passivated HEMTs. The percentage current losses were calculated from the 500 kHz data. ................... 89 7 1 Summary of IG at VG = 10V, Schottky barrier height before and after 5, 10 and 15 MeV proton irradiation with a dose of 5 1015 c m2 .............................. 158 7 2 Summary of the dependence of ON/OFF ratio, saturation drain current, subthreshold drain leakage current and subthreshold slope on proton irradiations. ....................................................................................................... 158 7 3 Summary of the dependence of normalized mobility, carrier concentration, sheet carrier concentration and carrier removal rate on the proton irradiations. ....................................................................................................... 159 7 4 Summary of threshold voltage shift, the reduction of extrinsic transconductance, sheet carrier concentration and mobility, as well as carrier removal rate of HEMTs prior to and post 5MeV proton irradiation with various doses. ............................................................................................................... 159 7 5 Summary of IDSS, sheet carrier concentration and mobility, as well as carrier removal rate of HEMTs as a function of proton irradiation doses. .................... 160 7 6 Summary of the dependence of drain breakdown voltage (VBR) and critical voltage (Vcri) during the off state drainvoltage stepstress as a function of irradiation dose. ................................................................................................ 160

PAGE 10

10 LIST OF FIGURES Figure page 1 1 The variation of the energy bandgap with corresponding lattice constant for the Group IIIV. ................................................................................................... 30 1 2 The schematic of a ty pical AlGaN/GaN HEMT device ........................................ 31 1 3 The conduction band in the AlGaN/GaN heterostructure. .................................. 32 1 4 The schematic of device operation ..................................................................... 33 2 2 Two dimentional simulations of the electrostatic field distribution between source and drain contacts for HEMT with and without source field plate ............ 43 2 3 Drain I V characteristics of HEMT with and without the source field plate. The VG starts f rom 0V and with a step of +0.5V ........................................................ 44 2 4 Off state gate currents as a function of VGS for HEMT with and without source field plate. The devices were stressed for 60 seconds at each gate voltage step, while grounding the source el ectrode and maintaining +5 V to the drain. The stress started at the gate voltage of 5 V, and the voltage step was kept at 1 V. ................................................................................................. 45 2 5 Off state total gate current (IG), gateto source current (IGS) and gateto drain current (IGD) as a function of VGS for HEMT with source field plate. .................... 46 2 6 Cross section TEM micr ographs of gate contact of HEMT The two white arrows indicate regions of Ni and oxygen diffusion and an associated threading dislocation. .......................................................................................... 47 2 7 TEM image of the region with Ni and oxygen diffusion and the associated threading dislocation. .......................................................................................... 48 2 8 The results of EE LS analysis ............................................................................. 49 3 1 Off state gate currents as a function of gate voltage for the HEMTs fabricated wi th d i f f e r e n t gate metallization. ......................................................................... 56 3 2 Drain IVs of HEMTs fabricated with d i f f e r e n t gate metallization measured before and after the off state stress ................................................................... 58 3 3 Schottky gate characteristics of the HEMTs fabricated with d i f f e r e n t gate metallization measured before and after the off state stress .............................. 59 3 4 Subthreshold drain current of AlGaN/GaN HEMTs fabricate d with d i f f e r e n t metallization bi ased at different drain voltage. .................................................... 60

PAGE 11

11 4 1 DC characteristics and electrical s tress data ...................................................... 69 4 2 Slope of subthreshold drain current of the AlGaN/GaN HEMTs before and after off state drain bias voltage stepstress as a function of the ambient temperature. ....................................................................................................... 70 4 3 Sub threshold drain leakage current and reverse bias gate leakage current for reference HEMT measured at the temperatures of 300 and 573K. ............... 71 4 4 Device isolation current at a drain bias voltag e of 10V and the gate leakage current at a gate voltage of 10V. ....................................................................... 72 4 5 The results of trap characteristi cs and device isolation current ......................... 73 5 1 Wafer scale m aps fo r unpassivated HEMTs. ...................................................... 81 5 2 DC characteristics of un passivated and passivated HEMTs ............................. 82 5 3 IDSVDS characteristics of u n p a ssi v a t e d a n d p a ssi v a t e d d e v i ce s. ....................... 83 5 4 DC and RF characteristics of HEMTs with different passivation layers .............. 84 5 5 Drain pulse measurements on unpassivated and passivated AlGaN/GaN HEMTs ............................................................................................................... 85 5 6 Gate pulse measurements on unpassivated and passivated AlGaN/GaN HEMTs ............................................................................................................... 86 5 7 Double pulse measurements with drain pulsed from 10V to 5V while simultaneously pulsing the gate from 6V to the value shown on the x axis ...... 88 6 1 Schematic of epitaxial structures of AlGaN/GaN HEMTs. .................................. 99 6 2 Transfer characteristics of HEMTs fabricated on different buffer layers. .......... 100 6 3 Sub threshold drain I V characteristics of HEMTs fabricate d on different buffer layers. ..................................................................................................... 101 6 4 Drain characteristics of HEMTs fabricated on different buffer layers. ............... 102 6 5 Gate characteristi cs of HEMTs fabricated on different buffer layers. ................ 103 6 6 Off state drain stepstress results of HEMTs fabricated on different buffer layers, measurements were conducted at VG = 6V. ........................................ 104 6 7 Some of the saturation drain I Vs recorded between each step stress for HEMTs fabricated on different buffer layers. .................................................... 105

PAGE 12

12 6 8 The results of off state breakdown voltage and isolation breakdown voltage measurements of HEMTs fabric ated on different buffer layers ......................... 106 6 9 Normalized drain current, IDS, as a function of VGS for b oth pulsed and dc modes. .............................................................................................................. 107 6 10 Room temperature photoluminescence (PL) spectra of HMETs with different buffer layers. Insert: Enlarged blue luminescence (BL) bands of those HEMTs with different buffer layers. ................................................................... 108 7 1 Off state gate currents as a function of drain voltage for the unirradiated and protonirradiated HEMTs. The protons had energy of 5 and 10 MeV and dose of 51015 cm2. .................................................................................................. 134 7 2 Drain I Vs of un irradiated HEMTs prior to and after off state electrical stepstress. The devices were stressed with VG = 6 V for 60 s at each drain voltage step until sudden increase o f IG was observed. .................................... 135 7 3 Drain I Vs of HEMTs irradiated with 5 MeV and 51015 cm2 doses of protons prior to and after off state electrical stepstress. The devices were stressed with VG = 6 V for 60 s at each drain voltage step until drain voltage reached +100V. .............................................................................................................. 136 7 4 Gate characteristics of unirradiated and protonirradiated (with 5 MeV and 51015 cm2 dose) HEMTs prior to and af ter off state electrical step stress. Un irradiated devices were stressed with VG = 6 V for 60 s at each drain voltage step until sudden increase of IG was observed. The same condition used for irradiated HEMT except the drain voltage reached +100 V. ............... 137 7 5 Off state breakdown measurement result of unirradiated and proton irradiated HEMTs. The incident protons had energy of 5 MeV and dose of 51015 cm2. ...................................................................................................... 138 7 6 A schematic of the AlGaN/GaN high electron mobility transistors. ................... 139 7 7 SRIM simulation results at the AlGaN/GaN interface of the AlGaN/GaN HEMT structure ............................................................................................... 140 7 8 D C ch a r a ct e r i st i cs of HEMTs pre and post proton irradiation with energy of 10 MeV at a fluence of 5 1015 cm2 ................................................................ 141 7 9 Gate IV of HEMTs pre and post proton irradiation with an energy of 10 MeV at a fluence of 5 1015 cm2. ............................................................................ 142 7 10 Off state drainvoltage a n d o ff state drain breakdown voltage of HEMTs p re a nd post proton irradiation. ............................................................................... 143

PAGE 13

13 7 11 IDS and IG as a function of VG of HEMTs pre and post proton irradiation with an energy of 10 MeV at a fluence of 5 1015 cm2. .......................................... 144 7 12 Drain I Vs of HEMTs pre and post proton irradiation with an energy of 10 MeV ................................................................................................................. 145 7 13 C arrier removal rate as a function of proton irradiation energies at a fixed fluence of 5 1015 cm2 .................................................................................... 146 7 14 T h e r e su l t o f G a t e p u l se m e a su r e m e n t . .......................................................... 147 7 15 T h e r e su l t o f d o u b l e p u l se m e a su r e m e n t ........................................................ 148 7 16 T h e r e su t l s o f s m a l l si g n a l m e a s u r e m e n t ........................................................ 149 7 17 T h e r e su t l s o f large si g n a l m e a s u r e m e n t ......................................................... 150 7 18 Percent increases of sheet resistance (RS), contact resistivity (RC), and transfer resistance (RT) after 5MeV proton irradiation with different doses. ...... 151 7 19 T h e r e su l t o f S R I M si m u l a t i o n .......................................................................... 152 7 20 D C ch a r a ct e r i st i cs of HEMTs before and after 5 MeV proton irradiation. ......... 153 7 21 Drain characteristics of HEMTs prior to and post 5MeV proton irradiation with various doses. .................................................................................................. 154 7 22 T h e r e su l t s o f Gate pulse measurements ........................................................ 155 7 23 T h e r e su l t s o f Drain pulse measurements ....................................................... 156 7 24 Off state drain stepstress of HEMTs prior to and post 5MeV proton irradiation with various doses. .......................................................................... 158 8 1 Schematic of the excimer laser drilling system. ................................................ 169 8 2 T h e d r i l l ing r a t e o f S i C ..................................................................................... 170 8 3 Photograph of a SiC sample drilled with four different sizes of circular via ....................................................... 171 8 4 Cross sectional view of SEM images of via holes with d i f f e r e n t diameter s. ..... 172 8 5 Photograph of arrays of rectangular via holes with the dimension of (column 100 100 ....................... 173

PAGE 14

14 8 6 Cross sectional view of SEM imagines of rectangular via holes with 90100 100 100 100 .......................................................... 174 8 7 The cr o ssse ct i o n a l v i e w o f depth around 90 ......................................................................................... 175

PAGE 15

15 LIST OF ABBREVIATIONS 2DEG 2 dimensional electron gas ALD Atomic layer deposition AlGaN Aluminum gallium nitride AlN Aluminum nitride C V Capacitance voltage CDLTS C onductance deep level transient spectroscopy DLTS D eeplevel transient spectroscopy EDX E nergy dispersive X ray EELS Electron energy loss spectroscopy FET Field effect transistor FIB Focused ion beam HEMT High electron mobility t ransistor ICP I nductively coupled plasma InAlN Indium aluminum nitride GaN Gallium nitride LEDs Light emitting diodes MESFETs Metal semiconductor field effect transistors MOCVD Metal Organic Chemical Vapor Deposition MOSFETs Metal oxide semiconductor field effect transistors MBE Molecular Beam Epitaxy PAE Power added efficiency PECVD Plasma enhanced chemical vapor deposition PL Photoluminescence Rc Contact resistivity

PAGE 16

16 Rs Sheet resistance RT Transfer resistance SBH Schottky barrier height SEM Scanning electr on microscopy SiC Silicon Carbide SRIM Stopping and range of ions in matter TDs Threading dislocations TEM Transmission electron microscopy TLM Transmission line measurement

PAGE 17

17 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 RELIABILITY STU D Y OF GAN BASED HIGH ELECTRON MOBILITY TRANSISTORS By Lu Liu December 2013 Chair: Fan Ren Major: Chemical Engineering Compound semiconductors are III V semiconductor nitrides, such as AlN, GaN and AlGaN, which have attracted plenty of attention due to their extraordinary material properties, such as high mobility, high saturation velocity and power density, and become a promising alternative for microwave power device. They have demonstrated excellent performance in such applications as UV detectors, UV and visible light emitting diodes (LEDs), microwave power amplifications, satellite, radar and wireless communication systems. Ho wever, the main impediment of this technology is lack of reliability. In order to improve the reliability of GaN HEMTs technology, it is essentially important to understand the nature of device degradation. In this dissertation, the effects of source field plates, gate metallization, passivation layers, buffer structures and proton irradiation on the dc/rf performance and reliability of AlGaN/GaN high electron mobility transistors (HEMTs) under off state stress conditions were investigated using stepstress cycling. The source field plate alleviated the peak electrical at drain side of gate edge and enhanced the drain breakdown voltage from 55V to 155V and the critical voltage (Vcri)

PAGE 18

18 for off state gate stress from 40V to 65V, relative to devices without the field plate. The critical voltage of electrical stepstress was increased more than 100% for HEMTs with Pt/Ti/Au gate metallization as compared to HEMTs with Ni/Au gate metallization. As compared to Al2O3 and HfO2, SiNx was found to be the most effective in reducing current collapse and also reduced fT and fMax through additional parasitic capacitance. The HEMTs with the thick GaN buffer layer showed the lowest critical voltage (Vcri) during off state drain stepstress, but this was increased by around 50 and 100% for devices with the composite AlGaN/GaN buffer layers or thinner GaN buffers, respectively. The dependence of dc and rf characteristics and reliability of GaN based HEMTs on proton irradiation energies and doses were examined. GaN based HEMTs sho wed a remarkable resistance to high energy protoninduced degradation and improved device reliability. In addition, temperature dependent subthreshold slope measurement was developed to study the effect of off state electrical stress on the trap densities and two traps with different activation energies at temperature range of 300 493K and 493 573K were identified. Finally, The laser micromachining of SiC by 193nm ArF excimer laser produced much higher etch rates (229870 m/min) than conventional dry etch ing (0.21.3 m/min) and the via entry can be tapered to facilitate subsequent metallization.

PAGE 19

19 CHAPTER 1 INTRODUCTION 1.1 Introduction to GaN based High Electron Mobility Transistors (HEMTs) There is increasing need for high power, high efficiency and linearity amplifiers for communication system s and defense radar. III nitride materials with wide bandgap are promising candidates for these applications. Examples are AlN and GaN with direct bandgaps of 3.4 and 6.2, respectively. The variation of the energy bandgap with corresponding lattice constant for the Group IIIV is shown in Figure 11. Gallium Nitride (GaN) based highelectronmobility transistors (HEMTs) are the next generation of power transistor technology, which was first demonstrated in 19931 and have exhibited impressive attributes and outstanding potential for high voltage and high switching frequency applications.2 6 In the pas t few decades, some aspects including material quality, choice of substrate, epi layer structures, advanced device designs and processing techniques have been studied intens iv ely. Recently, AlGaN/GaN HEMTs have attracted plenty of attention due to its extr aordinary material properties, such as high mobility, high saturation velocity and power density, and have therefore become a promising alternative for microwave power devices.7 10 As a result of the large conduction band discontinuity between Al0.25GaN (~4.2 eV) and GaN (3.4 eV), high electron mobility and high electron saturation velocity have been demonstrated in AlGaN/GaN material system s. Electron mobility ex ceeds 1500 cm2/V s and saturation velocity is around 2.5107 cm/s, respectively, and are beneficial for high frequency operation. Thus far, a current gain cutoff frequency (fT) of 225 GHz with a gate length (LG) of 55 nm11 and a power gain cutoff frequency (fmax) of 300 GHz with a gate length of 60 nm12 have been demonstrated in AlGaN/GaN HEMTs. In addition, the presence of

PAGE 20

20 strong piezoelectric effect and spontaneous polarization in III nitride leads to a high carrier concentration (higher than 1013 cm2). As a result, a high current density can be achieved in AlGaN/GaN heterostructures without conventional doping.13 14 Moreover, GaN has a large band gap of 3.4 eV and a high breakdown electric field (~3 MV/cm), which make s it capable of handling high voltage. A record breakdown voltage of 2200V with a LGD 15 The high current density and high breakdown voltage of this martial make it an excellent candidate for high power applications. To date, record output power densities of 5.1W/ mm at 31 GHz,16 16.7 W/mm at 10 GHz and 20.7 W/ mm at 4 GHz have been reported.17 Furthermore, the good thermal stability of GaN based HEMTs allows operation at high temperatures.18 1.2 Motivation In spite of the extraordinary material properties of AlGaN/GaN heterostructure system s, such as high mobility, high saturation velocity and high breakdown field, the main impediment to this technology is lack of reliability. During device operation, the main degradation modes observed in previous work are increasing of gate leakage as well as subthreshold drain leakage, and the worsening of traprelated effects, which result in a reduction in drain current. Recently, some degradation mechanisms have been proposed. They are hot electroninduced trap degrad ation19, crystallographic defect formation through the inverse piezoelectric effect2022 and electric field driven mechanism under off state conditions.23 1.3 Background 1.3.1 GaN based HEMTs A transistor is a semiconductor device containing three or more terminals, the main func tions of which are to switch electronic signals, or to amplify, owing to the fact

PAGE 21

21 that the controlled (output) power can be higher than the controlling (input) power. In a Field Effect Transistor (FET), there are three terminals, including source (S), thro ugh which the carriers enter the channel, Drain (D), through which the carriers leave the channel and Gate (G) the terminal that modulates the channel conductivity. Conventionally, current entering the channel at S is designated by IS, current entering the channel at D is designated by ID, Drain to source voltage is VDS. By applying voltage to G, one can control ID.24 The device consists of an active channel through which charge carriers, electrons or holes, flow from the source to the drain. Source and drain terminal conductors are connected to the semiconductor through ohmic contacts, and the gate terminal through Schottky contact. The conductivity of the chan nel is a function of the potential applied across the gate and source terminals. The gate can be separated from the channel by an insulator (as in a MOSFET), can form a pn junction (JFET), or a Schottky barrier junction with the channel [Metal Semiconduct or FET (MESFET)]. The High Electron Mobility Transistor (HEMT) is a modification of FET, also known as as heterostructure FET ( HFET ), which is a FET incorporating a junction between two materials with different band gaps as the channel instead of utilizing a doped region. Currently, commonly used material combinations are AlGaN, InAlN and AlN with GaN. The schematic of a typical AlGaN/GaN HEMT device is shown in Figure 12. In a HEMT, the current flows between the source (S) and drain (D) terminals (ohmic c ontacts) through a channel, which is called a two dimensional electron gas (2DEG) channel. The channel conductance is modulated by an electric field produced by the voltage between the source and gate.

PAGE 22

22 The 2 DEG channel is located inside the smaller bandgap material (GaN (3.4 eV) < Al0.25GaN (~4.2 eV)) and near the heterostructure interface. 1.3.2 Origin of 2DEG Figure 13 shows the conduction band in the AlGaN/GaN heterostructure. The dash ed line indicates the position of the Fermi level in the heterostr b is the C is the conduction band offset, and E0 is the lowest subband level of the 2DEG. The labels correspond to the ones used i n equation1 and equation225. The large polarization difference at the heterostructure interface induces a high electron sheet density, ns. Th e electron density is affected by the barrier height b and the electron sheet density is calculated as13 26 n ( x d ) = ( x ) e ( x ) e d ( e ( x ) + ( x ) E ( x ) ) (1 1) the AlGaN/GaN interface, C is the conduction band offset, 0 is the dielectric constant of vacuum, and is the using the expression 1.3.3 The operation of FET At small drain source voltage VD, the drain c urrent ID linearly increases with VD. As shown in Figure 14 A a fully open channel is present when a positive VG is applied ( x ) = ( 9 e 8 8 m n ( x d ) ( x ) ) + m n ( x d ) (1 2 )

PAGE 23

23 when a negative voltage is applied to the gate, t he electrons in the channel are partially depleted, and the resistance of the con ducting channel increases, as shown in Figure 14 B As a negative gate voltage VG increased, a threshold voltage Vth is reached. At the threshold, the electrons in the 2DEG channel are completely d epleted and no electrons flow through the channel (the chan nel is closed). This condition is called pinchoff, as shown in Figure 14 C The transistor is switched on (the channel is conductive) at zero gateto source voltage ( VG) and requires a negative gateto source voltage to cut off the channel current (IDS). This is known as a depletion mode or normally on transistor Conversely, if a transistor, which is off at zero VG, and requires a positive gate voltage to switch the device on is known as an enhancement mode or normally off transistor 1.3.4 Substrates One of the key challenges in the development of GaN HEMT tec hnology is the selection of s ubstrate materials which provide highquality GaN epitaxy with low density of impurities and excellent thermal conductivity in order to dissipate heat generated duri ng device operation. T hree substrates that have been widely used as substrates in the GaN technology, including Silicon (Si), sapphire (Al2O3), and silicon carbide (SiC). Recently, diamond and GaN have been employed as substrates as well. The advantage of the diamond is its high thermal conductivity, while the GaN offers a low density of impurities. Table 11 summarizes the properties and challenges for Al2O3, SiC, Si, and GaN substrates .27 1.3.4.1 Silicon (Si) Si has a reasonable cost and an acceptable thermal conductivity which is similar to bulk crystalline GaN, but is inferior to that of SiC and sapphire. In order to achieve the growth of high quality GaN on S i (111), the lattice misfit (~17%) and thermal expansion

PAGE 24

24 coefficient (TEC) mismatch between Si and GaN must be overcome. In addition, the thermal stability of Si at typical GaN growth temperature is also inferior to SiC and sapphire. GaN grows on Si substr ate with a tensile stress since Si has a larger lattice constant than GaN, which leads to the creation of crystal defects, degrading the performance of the device. On order to overcome the impact of the inherent properties in Si on epitaxial crystal qualit y, a nucleation layer is applied to ensure good GaN onSi crystal quality.2831 1.3.4.2 Silicon Carbide (SiC) The lattice mismatch between SiC and GaN (~4%) is re latively minor, and allows for the formation of high quality GaN onSiC structure s than can be achieved on either GaN onSi or GaN onAl2O3 counterparts. T he density of dislocations in GaN layers grown on SiC substrate is under 3108 cm2. Among all substr ates aforementioned, SiC is preferred for high power and high frequency applications. However, it is expensive compared to Si or sapphire and its application in the field is limited due to cost consideration. 1.3.4.3 Sapphire (Al2O3) Sapphire is cheap and available in wafers with large diameters, so it is one of the more commonly employed substrates However, it has the largest lattice misfit (~14%) and thermal expansion coefficient mismatch with GaN ( the thermal mismatch varies from 14 to 26%, depending on the their relative orientation of the crystals) Another main drawback of this substrate is its poor thermal conductivity, which severely limits its use in applications where efficient dissipation of heat is required. The exceptionally high current in 2DE G channel of GaN power devices produces significant amount of heat during device operation. P oor thermal conduct ivity can cause device overheating and

PAGE 25

25 degrade the device performance, even during I V sweeps. With the development of GaN epitaxy technology, G aN onsapphire is being replaced by GaN onSi and GaN onSiC for HEMTs fabrication. 1. 4 GaN Reliability Issues 1. 4 .1 Hot electron Effect One well known failure mechanis m of early GaAs based devices was hot electroninduced degradation.19, 32, 33 Under higher drain bias, the accelerated electrons gain a great deal of enough energy in the channel and may be trap ped one the device surface, in the AlGaN barrier layer or in the GaN buffer layer. This can lead to a shift in threshold voltage an increase in drain resistance and a reduction in drain saturation current.6 34 Hot electron induced degradation has been observed in AlGaN/GaN HEMTs with SiNx and SiO2 passivatio n on sapphire and SiC substrates under dc and rf stress conditions With high drain bias and high input drive, hot electrons are created in the region betw een the gate and drain, and generate permanent traps which cause increased surface depletion, increas ed series resistance, and reduced gat e drain electric field. Not only can hot electrons be trapped but they can also induce trap formation at surface region between gate and drain. In a 3000hour test on AlGaN/GaN HEMT under onstate and off state conditions, a decrease of the drain current and transconductance, and an increase of the channel resistance were observed by Sozza et al. Through the use of low frequency techniques, they were able to attribute the increase in traps density at the surface between the gate and drain to hot electrons present in the channel.35 To avoid degradation by hot electrons caused by the presence of high electric field in the g ate drain region, proper solutions are needed such as by adopting surface passivation and field plate.

PAGE 26

26 To investigate hot electron effect, electroluminescence (EL) was conducted at different gate and drain voltages. The information of the amount and energetic distribution of hot electrons can be obtained from EL characterization.19, 36 1. 4 .2 Inverse Piezoelectric Effect Recently, another degradation mechanism was proposed by Joh and del Alamo,21, 37 crystallographic defects formed through inverse piezoelectric effect in GaN based HEMTs Due to the lattice mismatch between AlGaN and GaN, there is a great deal of elastic energy stored in AlGaN/GaN heterostructure even in absence of bias. Typically, AlGaN on GaN is under tensile strain.38 When tensile mechanical strain was applied on devices intentionally during electrical stress, a 13V lower value of critical voltage was found compared to control samples.22 In addition, induced piezoelectric strain/stress in gate drain region under different bias conditions was evaluated by microRaman measurement.39 During device operation under high drain bias, a large electric field appears around the drainside of the gate edge across barrier layer This induces a large amount of mechanical stress, which would trigger the formation of crystallographic defects and relaxation of the strain when the electric field reaches a certa in value. Those defects behave as traps, which degrade carrier transport properties and carrier concentration in the channel, resulting in increasing of drain resistance and decreasing of saturation drain current and transconductance. The gate leakage curr ent is increased as well through trapassisted tunneling in the barrier layer. Cross sectional transmission electron microscope (TEM) results revealed the formation of pits or cracks in the AlGaN barrier layer close to gate edges.40 The degree of physical damage was correlated to the reduction in drain current. As confirmed by TEM, higher degradation in drain current, resulted in the creation of more severe physical defects Energy dispersive X ray (EDX)

PAGE 27

27 measurement also found that SiNx got into formed pits or cracks and oxidation happened during device degradation .40 Those results also suggested the reduced strain between barrier layer and buffer layer would improve device reliability. InAlN with an In mole fraction of 0.17 can be grown latticematched to GaN, which eliminates the strain present in the AlGaN /GaN heterostructure system and should be beneficial to device reliability.21 1. 4 .3 Trapping Effect Another important parameter in the evaluation of device reliability is current collapse, the reduction of drain current after the application of high voltage. This is caused by the trapping effect. This effect limits the performance of HEMTs in high power and high frequency applications. The origin of current collapse was attributed to surface trapping, in the AlGa N barrier layer, or buffer trapping, in the GaN buffer layer,4 1 42 leading to an increase in dispersion between direct current (dc) and pulsed current voltage (I V) characteristics, and a decrease in maximum transconductance and saturation drain current. Due to the strong polarization of the material, surface states are unavoidable.43 44 The donor like traps capture electrons tunneling from the gate metal and become neutral. F urther accumulation of elec trons at the surface would make the surface potential negative and form an extended depletion region, a socalled virtua l gate, which can deplete electrons in the channel, resulting in a reduction in the drain saturation current reduces.45 To determine a detailed degradation mechanism, it is very important to understand the trapping behavior, such as the phy sical locations, energy levels and trapping/detrapping time constants. So far, a great deal of study has been conducted toward the achievement of that goal. Various trap characterization methods, such as

PAGE 28

28 deeplevel transient spectroscopy (DLTS),4648 frequency dependent capacitance and conductance measurements,4951 and capacitance voltage (C V)52, 53 measurements were employed to study trapping effects. The activation energies of detected traps are range from 0.07 to 0.84 eV .46 49 The time constant varies between an order of micro second s to millisecond s for fast and slow traps, respectively.49, 50 According to report s, trap densiti es are generally around 1012 cm2V1 and corresponding traps are located either in the AlGaN barrier layer or at the AlGaN/GaN interface.4951, 53 It is well known that surface passivation, such as SiNx, is an effective way to reduce trapping effects.5458 In addition, it b enefits devices pinch off characteristics and reduces gate leakage current.55 The addition of a SiNx passivation layer to undoped AlGaN/GaN HEMTs also increases the saturated power density by up to 100% at 4 GHz and increased the breakdown voltage by an average value of 25%.54 Compared to unpassivated HEMTs which suffer severe ID collapse (96%), passivation suppressed more than 80% of the ID collapse. As opposed to surface trapping, bulk/buffer trapping a ctually reduces ID in the case of SiNx passivated HEMTs.58 P ossible mechanisms o f the suppression of trapping effects by SiNx passivation are 1) passivation prevents the electrons from tunneling from the gate metal reaching surface states 2) Replacement of surface donors by the incorporation of passivation between Si and surface states.45 1. 5 Dissertation Outline This dissertation covers three main topics, including the reliability study of GaN based high electron mobility transistors (HEMTs), the radiation effects in GaN based HEMTs and the micromachinin g of Silicon Ca rbide (SiC). B ackground knowledge on the III V compound semiconductor transistors, especially, the properties and current status of gallium nitride (GaN) based high electron

PAGE 29

29 mobility transistors, as well as the reliability issues hindering the further dev elopment of GaN HEMTs technology are presented in Chapter 1. The next several chapters cover intensive research studies on the topics of reliability issue s of GaN HEMTs outlined in the Chapter 1, which were performed in the past few years at the Universit y of Florida. Chapter 2 discusses the effect of the source field plate on the dc characteristics and reliability of AlGaN/GaN HEMTs under off state electrical step stress. Chapter 3 covers the improvement of dc characteristics and reliability under off sta te electrical stepstress of AlGaN/GaN HEMTs which were achieved by employing platinum gate metallization. A method to estimate the trap density of AlGaN/GaN HEMTs by temperature dependent subthreshold slope measurement is presented in the Chapter 4, and the effect of off state electrical stress on the trap density is studied. In Chapter 5, the passivation properties, including current collapse and rf performance, of HfO2, SiNx and Al2O3 on AlGaN/GaN HEMTs are compared. Chapter 6 investigates the effect of buffer structure on the reliability of AlGaN/GaN HEMTs. Chapter 7 focuses on the radiation effects in the GaN based H EMTs, which includes the study of the reliability of protonirradiated InAlN/GaN and the effects of proton energy and dose on the dc characteristics and reliability of AlGaN/GaN HEMTs. Chapter 8 reports the improvement of drilling rate and via holes profile on SiC using ArF excimer laser. Chapter 9 provides a summary and co nclusion for all the topics discussed in this dissertation.

PAGE 30

30 Figure 11. The variation of the energy bandgap with corresponding lattice constant for the Group IIIV.

PAGE 31

31 Figure 12. The schematic of a typical AlGaN/GaN HEMT device

PAGE 32

32 Figure 13. The conduction band in the AlGaN/GaN heterostructure.

PAGE 33

33 Figure 14. The schematic of device operation, A) A fully open channel presents when a positive VG is applied. B ) Conduction band of a HEMT with negative applied VG, but the c hannel is still open. C ) The negative VG is larger than the threshold voltage Vth and the channel is closed.

PAGE 34

34 Table 11. Summary of the properties of Al2O3, SiC, Si, and GaN and the challenges. Substrate Lattice match (%) TEC mismatch (%) Thermal Condu ctivity (W/cm K at 25 o C) Substrate size (mm) Substrate cost (relative) Integratability Si 17 56 1.5 400 Low Very high SiC 3.5 25 3.0 3.8 150 Very high Moderate Al 2 O 3 14 26 0.4 150 High Low

PAGE 35

35 CHAPTER 2 THE EFFECTS OF SOURCE FIELD PLATE ON THE CHARA TERISTICS OF OFFSTATE, STEPSTRESSED ALGAN/GAN HIGH ELECTRON MOBILITY TRANSISTORS 2.1 Introduction to Field Plate The promising performance of AlGaN/GaN high electron mobility transistors (HEMTs) under high frequency and high output power density conditi ons has intensified efforts to understand reliability and degradation mechanisms under different operating conditions .1921, 37, 5967 At high source drain biases, crystallographic defects and even cracking can occur as a result of t he inverse piezoelectric effect .20, 21 37, 5961 Under these conditions, the presence of strong electrostatic fields in the piezoelectric GaN and AlGaN layers leads to additional mechanical stress that is concentrated in the AlGaN barrier layer. At high enough field, the change in AlGaN elastic energy can produce extended and point defects. The degradation under off state conditions is electric field driven, with devices of different gate length failing at different drainsource biases but similar electrostatic field thresholds Field plates have been employed for AlGaN/GaN HEMT and GaAs met al semiconductor field effect transistors (MESFETs) to tailor the electric field profile near the drain edge of the gate for enhancing the breakdown voltage and also reducing the effect of current collapse.6870 It is well recognized that the field plate can improve th e gate and drain breakdown voltage, which would improve device reliability. Gate field plates and source field plates are the most common types used for power applications. The gate field plate reduces the peak electric field at the edge of the gate on the drain side. However, the gate field plate increases gateto drain capacitance, reduces the saturation current and degrades the rf gain characteristics. The source field plate can

PAGE 36

36 increase the gate and drain breakdown voltage and reduce the gateto drain capacitance, and thus improve rf performance.71 In this chapter, the degradation of AlGaN/GaN HEMTs with and without source field plate under stepstressing of the gate reverse bias was reported. The devices exhibit sev eral threshold gate biases for the onset of increased gate leakage. The degradation of gate current is irreversible and is accompanied by a small decrease in drainsource current. Transmission electron microscopy (TEM) imaging and small probe microanalysis were used to examine the degradation occurring at the metal AlGaN interface. 2.2 Experimental 2.2.1 Material Growth The HEMT on Si wafers were grown by metal organic chemical vapor deposition (MOCVD) with conventional precursors in a coldwall, rotatingdisc reactor designed from flow dynamic simulations. The growth process was nucleated with an AlN layer to avoid unwanted GaSi interactions. The epitaxial stack consisted of a proprietary AlGaN transition layer ,72, 73 ~800nm GaN buffer layer, and 16nm unintentional doped Al0.26Ga0.74N barrier layer. The nominal growth temperature for the GaN buffer and AlGaN barrier layers was 1030C. 2.2.2 Material Growth and HEMTs Fabrication HEMT fabrication began with Ti/Al/N i/Au Ohmic metallization and rapid thermal annealing in flowing N2 at approximately 825 C. Contact resistance, specific contact resistivity, and specific onmm, 5 106 cm2mm, respectively. Inter device isolation was ac complished by use of multiple energy N+ implantation to produce significant lattice damage throughout the thickness of the GaN

PAGE 37

37 buffer layer. The ion implantation step maintains a planar geometry in the fabricated device and reduces parasitic leakage paths that may exist in pas sivated, mesaisolated HFETs .74 Immediately following implantation, the wafers were passivated with 70nm thick SiNx in a PECVD chamber maintained at a base plate temperature of 300C. Schottky gate definition was achieved by patterning the gate and selectively removing the SiNx passivation layer. The contact windows to the Ohmic contact pads were also opened at the same time as defining the Schottky gate. After SiNx etching, wider gate patterns were redefined with another photolithography step and contact windows to the Ohmic contact pads were also opened. Ni/Au based gate metallization was deposited on the gate and Ohmic contact pads simultaneously. The wafers were then passivated with another 400nm layer of PECVD SiNx at 300C. The contact windows were opened by dry etching. There was an additional metal deposition for the HEMT with the source field plate. The field plate was connected to the source terminal and extended by 1 m out over the gate to the gateto drain region. The source to gate distance and channel length of the HEMTs with and without the source field plate were kept constant at 1 and 4.7 m, respectively. A schematic of the HEMT with the source field plate, and optical micrographs of the HEMT with and without source field plate, are shown in Figure 21. 2.2.3 Electrical Step stress The HEMTs were step stressed biased in the dark at room temperature with up to 90V reverse gate voltage at a fixed sourcedrain bias of 5V using an HP 4156C semiconductor parameter analyzer. During the stress, t he devices were stressed for 60 seconds at each gate voltage step, while grounding the source electrode and maintaining +5 V to the drain. The stress started at the gate voltage of 5 V, and the voltage step was kept at 1 V. During the stepstress, gate to source leakage current,

PAGE 38

38 IGS, and gateto drain leakage current, IGD, were also measured. Between each s tep stress, drain I V, extrinsic transconductance, gate forward current biased from 0 to 1.5 V and gate reverse current biased from 0 to 5 V, source and drain resistance were recorded. Self heating effects were negligible based on the low drain source cur rents under our test conditions, as supported by thermal simulations. 2.2.4 Device Characterization and Simulation Cross sections of some devices were prepared for TEM examination using a Nova 200 focusedion beam system. Finally, the Atlas code from Silva co was used to simulate the electric field for HEMTs with and without source field plate. 2.3 Results and Discussion Figure 22 shows the 2D electric field simulation results of the electric field distribution around the source, gate, and drain contacts for HEMTs with and without source field plate. The simulator is physically based and involves solution of the drift/diffusion model using Poisson and carrier continuity equations, also taking into account carrier statistics, impact ionization, lifetime, mobi lity and generationrecombination. There were two peaks of the electric field at the gate edges, and the drain side of the gate edge exhibited the maximum electric field. The source field plate over the gate electrode reduced the peak electric field at the drain side of the gate edge: this would improve the off state breakdown voltage of the HEMTs. Figure 23 shows the drain I V characteristics of HEMTs with and without source field plate. Both devices revealed similar saturation current and threshold volt age. The device without the source field plate displayed an off state drain breakdown voltage of 55 V and a significant increase of the breakdown voltage to greater than 150 V was observed, as illustrated in the insert in Figure 23.

PAGE 39

39 Figure 24 shows the typical stepstress results of more than 30 HEMTs with and without the source field plate; the gate current, IG, has been plotted as a function of the stressed gate voltage. The critical voltage of the off state step stress was defined as the onset of IG i ncrease during the stress. It has been reported20 that once the gate voltage of the HEMTs reached a critical voltage, a decrease of saturation drain current and an increases of the source and the resistance were observed. Similar results were obtained in this experiment. The typical critical voltages of HEMTs without the source field plate were around 40 V. This value increased to around 65 V for the HEMTs with the source field plate. This increase of the critical voltage for the HEMT with the source field plate was att ributed to the reduced electric field on the drain side of the gate edge, consistent with the simulated electric field results. Above the critical voltage, there were additional increases of IG observed at high gate bias voltages, as shown in Figure 25. The total gate leakage current, IG, is the sum of the gateto source leakage current, IGS, and gateto drain leakage current, Igd. The first onset of IG increase matched with the increase of IGD. This result implied that degradation occurred on the gate edge close to the drain side when the gate bias voltage reached the critical value. This degradation phenomenon has been report ed previously by several groups .20, 21 37, 59, 60 For the second IG increase, there was no change for IGD, but an increase of IGS was observed. The amount of IGS increase corresponded to the increase of IG. Thus, it seems likely that some degradation also occurred on the gate edge close to the source side. There were some occasions when both IGS and IGD increased under high stress bias voltage.

PAGE 40

40 In order to establish whether t here were degradations on either the source or drain sides of the gate edge, TEM was used to examine the gate edges. As shown in Figure 26 A the metal / semiconduct or interface for the unstressed sample was very sharp, and no reactions between the gate metal contact and semiconductor were observed. A small void was present in the Ni, but this was most likely an artifact of the focused ion beam (FIB) milling. For the stressed sample, a notch was observed on both sides of the gate, as shown in Figure 26 B and C These defective regions could be the initial stages of cracking due to the inverse piezoelectric effect, as previously reported.19, 59, 62 Other regions of the stressed sample showed that Ni had interacted with the underlying nitride layer close to the position of an associated threading dislocation (TD), as further illustrated in Figure 27. Electron energy loss spectroscopy (EELS) line scans were performed across the gate/AlGaN/GaN heterostructure around the area of the diffused gate metal, as shown in Figu re 2 8 D Line scan a was made vertically across a pristine interface and no metal diffusion was observed. A thin oxide layer was, however, observed at the metal/semiconductor interface and metal/oxide/semiconductor interfaces was very abrupt. The line sc an b was made vertically across the region of metal diffusion. The dashed lines across Figure 28 A and B indicate the interface between the original metal and the AlGaN layer. The scans clearly show that both Ni and O were diffused about 34 nm into the AlGaN layer at the site of the TD. As shown by the horizontal EELS line scan c, Figure 28 C It also appears that the Ni had diffused laterally. Such Ni diffusion regions are likely to be additional sources responsible for the observed increase of the gate leakage current.

PAGE 41

41 2.4 Summary In this chapter, the effect of source field plates on the reliability of AlGaN/GaN HEMTs under off state stress was present. With the incorporation of the source field plate, critical voltages of ~65 V were achieved as compared with 40V for HEMTs without the source field plate. These results were consistent with electric field simulations and drain I V characteristics of HEMTs with and without source field plate; slightly lower peak electric field appeared at the gate edges f or the HEMTs and significant higher drain breakdown voltage for the HEMTs with the source field plate. The TEM observations revealed that pits attributed to the inverse piezoelectric effect had appeared on both the source and drain sides of the gate edges, and there was a thin oxide layer between the Ni and AlGaN layer. Nickel and oxygen diffusion were associated with a threading dislocation for the stressed HEMT, which would provide additional pathways for the increased gate leakage current.

PAGE 42

42 Figure 21. A ) Optical micrograph of a typical HEMT with source field plate. B ) Optical micrograph of a typical HEMT wit hout source field plate. C ) Schematic of HEMT with source field plate. 3 um 1 SiN x SiN x 1 um 2.5 L g = 440 nm Field Plate Source Drain Gat Field plate device (C ) (A) (B)

PAGE 43

43 Figure 22 Two dimentional simulations of the electrostatic field distribution between source and drain contacts for HEMT with and without source field plate. 0 1 2 3 4 5 6 7 8 9 0.0 0.5 1.0 1.5 2.0 2.5 3.0 FP NFPElectric Field (MV/cm)Position ( m )

PAGE 44

44 Figure 23. Drain I V characterist ics of HEMT: A) without; and B ) with, the source field plate. The VG starts from 0V and with a step of +0.5V (Insert) The off state drain breakdown voltage of HEMTs with and without field plate, which was measured at VG = 8V. 0 50 100 150 200 0 10 20 30 40 IDS (mA/mm) IDS (mA)VDS (V)VG = 0V Step = 0.5V 0 100 200 300 400 0 20 40 60 80 0 10 20 30 40 VG = 0V Step = 0.5VIDS (mA/mm) 0 50 100 150 200 0 4 8 FP NFPVG= -8VIDS ( A)VDS (V)IDS (mA)VDS (V)0 100 200 300 400 (A ) ( B )

PAGE 45

45 Figure 24. Off state gate currents as a function of VGS for HEMT with and without source field plate. The devices were stressed for 60 seconds at each gate voltage step, while grounding the source electrode and maintaining +5 V to the drain. The stress started at the gate voltage of 5 V, and the voltage step was kept at 1 V. 0 -20 -40 -60 -80 10-310-210-1100 Ig (A) Ig (mA/mm)VGS (V) FP NFP10-710-610-510-4

PAGE 46

46 Figure 25. Off state total gate current (IG), gate to source current (IGS) and gateto drain current (IGD) as a function of VGS for HEMT with source field plate. 0 -20 -40 -60 -80 10-410-310-210-1100 Ig (A) Ig (mA/mm)VGS (V) IG IGS IGD10-810-710-610-510-4

PAGE 47

47 Figure 26. Cross section TEM micrographs of gate contact of HEMT before stress A), and aft er stress B ) for the edge of the gate clos e to the source contact, C ) edge of the gate close to the drain contact. The two white arrows indicate regions of Ni and oxygen diffusion and an associated threading dislocation. 10 10 (A ) (B ) (C )

PAGE 48

48 Figure 27. TEM image of the region with Ni and oxygen diffusion and the associated threading dislocation. AlGaN GaN N i Au TD Metal diffusion

PAGE 49

49 Figure 2 8. The results of EELS analysis. A ) R elative compositions of Ni, O, and N distributions for a vertical EELS line scan acros s gate area without degradation, B ) relative compositions of Ni, O, and N distributions for a vertical EELS line scan across gate area with Ni and O diffusion. The dashed line across the topright and bottom right indicates the ori ginal metal and AlGa N interface, C ) relative compositions of Ni, O, and N distributions for a horizontal EELS line scan across the gate area with Ni and O diffusion at the diffused gate contacts. D ) TEM image of the region with Ni and oxygen diffusion and the associated threa ding dislocations, with the locations of EELS line scans as marked. 0 2 4 6 8 10 12 14 16 180 2 4 6 8 10 12 Counts ( 105)Distance (nm) N O Ni 0 2 4 6 8 10 12 14 16 18 0 2 4 6 8 10 12 Counts ( 105)Distance (nm) N O Ni 0 10 20 30 40 0 2 4 6 8 10 12 Counts ( 105)Distance (nm) N O Ni 10 nm Ni AlGaN A B C (A) ( B ) ( C ) (D)

PAGE 50

50 CHAPTER 3 THE EFFECT OF GATE M ETALLIZATION ON THE RELIABILITY OF ALGAN /GAN HIGH MOBILITY TRANSI STORS 3.1 Overview of Gate Metallization of AlGaN/GaN HEMTs AlGaN/GaN High Electron Mobili ty Transistors (HEMT) exhibit impressive attributes and outstanding high voltage swi tching and RF power performance .2 6 In spite of the extraordinary material properties of the AlGaN/GaN heterostructure system, such as high mobility, high saturation velocity and good thermal conductivity, the main impediment of this technology is lack of reliability. Recently, a number of degradation mechanisms have been proposed. Th ey are hot electronin duced trap degradation,19 crystallographic defect formation through t he inverse piezoelectric effect ,2022 electric field driven mecha nism under off state conditions ,20 gate sinking75 and Ohmic contact degradation.76 To date, Ni/Au gate metallization has been the most widely used in AlGaN/GaN HEMTs reliability studies. Pt contacts have shown promising characteristics for Schottky contacts on AlGaN/GaN, with a reported barrier height of 1.09 eV77 and improved thermal stability .78 Pt/Ti/Au contacts provide a barrier height of 1.18 0.07 eV on GaN and 2.0 0.1 eV on Al0.31Ga0.69N and exhibit lower leakage current and higher barrier height than that of Ni based contacts .79 The presence of an intermediate Ti layer, which attenuates the AuPt interaction, further improves t he t hermal stability of devices .7981 Pt/Ti/Pt/Au Ohmic contacts were used for AlGaAs/GaAs heterojunction bipolar transistors to achieve a lower contact resistance, due to the reaction of GaAs and Pt to form PtAs2.81 In addition, the control of threshold voltage to fabricate enhancement (E) mode HEMTs has been realized by using buried platinum gate technology in the InAlAs/I nGaAs/InP and AlGaN/ GaN systems.77

PAGE 51

51 In this chapter, AlGaN/GaN HEMTs with Ni/Au and Pt/Ti/Au gate metallization were fabricated on the same wafer and the drain and gate characteristics with Ni/Au and Pt/Ti/Au gate metallization before and after off state stress as well the drain current on/off ratio were compared. 3.2 Experimental 3.2.1 Material Growth and HEMTs Fabrication AlGaN/GaN heterostructures were grown on c plane sapphire by metal organic chemical vapor deposition. The epi layers consisted of a 1 m thick carbon doped GaN buffer, followed by a 55nm thick undoped GaN channel layer, 21 nm Al0.25Ga0.75N, and 2.2 nm GaN cap layer. The HEMT fabrication included Ohmic contact deposition by lift off of e beam deposited Ti/Al/Ni/Au, followed by rapid thermal annealing at 850C for 30 s in N2. A contact resistance of 0.6 mm was measured using the transmission line method. For device isolation, multiple doses and energies of N+ implantation were used to maintain a planar geometry and reduce parasit ic leakage current. Shipley Microposit STR 1045 positive photoresist was employed to protect the active region of the devices. Ni (200) /Au (800 ) and Pt (100 ) /Ti (200 ) /Au (800 ) Schottky gate metallization was defined by optical lithography and followed with standard lift off of the e beam deposited metals. The HEMTs were passivated using 400 nm of SiNx with a plasmaenhanced chemical vapor deposition (PECVD) system at 300C. The metal windows were opened with buffered HF. 3.2.2 HEMTs Characteri zation The device DC characteristics were measured with a HP 4156 parameter analyzer. The HEMTs off state step stresses were performed in the dark at room

PAGE 52

52 temperature with the gate biased up to 100V reverse gate voltage at a fixed source drain bias of 5V using the parameter analyzer. 3.3 Results and Discussion Figure 31A and B show the off state step stress results of more than 15 HEMTs with Ni/Au and Pt/Ti/Au gates. The gate current, IG, has been plotted as a function of the stressed gate voltage VGS. The Ni/Au gated devices were stressed for 60 seconds at each gate voltage step, while grounding the source electrode and maintaining constant +5 V on the drain. Some of the Pt gated devices were stressed at +5 V on the drain contact, as well as at higher drain voltages of 15 and 20V. The stress started at VGS = 10 V, and the voltage step was 1 V. During the stepstress, gate to source leakage current, IGS, and gateto drain leakage current, IGD, were also measured. Between each step stress, drain I V, ex trinsic transconductance, gate forward current biased from 0 to 1.5 V and gate reverse current biased from 0 to 5 V, source and drain resistance were recorded. Self heating effects were negligible based on the low drain source currents under our test conditions, as supported by thermal simulations. The critical voltage of the off state step stress was defined as the onset of IG increase during the stress. Typical critical voltages for electrical degradation of the Ni/Au get HEMTs ranged from ~45 65 V, as s hown in Figure 31 A During the off state stress, the electric field was highest at the gate edges. The presence of a strong electric field in the piezoelectric GaN and AlGaN led to additional mechanical stress concentrated in the AlGaN barrier layer. At a high enough field, the change in AlGaN elastic energy could produce extended and point defects. Thus it was suggested that the critical voltages are strongly dependent on electrical field. However, no such critical voltage was exhibited for Pt gated devic es fabricated on the same wafer, even the gate was biased to 100V. This

PAGE 53

53 implies something else also controlled the onset of the critical voltage. The thermal stability of Pt on GaN has been studied and Pt only reacted with GaN at the ambient temperature > 500C. Besides the electrical field, the thermal stability of the metal contact may play a role to the occurrence of a critical voltage. However, no such critical voltage was exhibited for Pt gated devices fabricated on the same wafer, even the gate was biased to 100 V. This implies something else also controlled the onset of the critical voltage. We previously reported that Ni and interfacial oxides diffused into GaN, threading dislocation generated under the gate contact and pits formed on the edges of the gate contacts for the Ni gated HEMT after reached the critical voltage via a transmission electron microscopy (TEM) study.82 All these defects could contribute the increase of the gate current. X ray photoelectron Spectroscopy (XPS) was also used to examine the thermal stability of the Ni and Pt metal contacts deposited on GaN.83 Ni reacted with the interfacial oxide for the HEMT sample annealed at 300C and no chemical interaction between the Pt and GaN. Pt only reacted with GaN at the ambient temperature >500C.78 Therefore, besides the electrical field, the chemical and thermal stability of the metal contact may play a role to the occurrence of a critical voltage. During the off state stress, before the gate bias voltage reaching the critical voltage, there was no degradation for both gate and drain IVs observed. Once the gate voltage of the HEMTs reaching the critical voltage, not only the gate reverse bias leakage current suddenly increased, as illustrated in Figure 31 A but also the saturation drai n current and the Schottky barrier height were permanently decreased, as shown in Figure 32 and Figure 33. This degradation was irreversible when the biased gate

PAGE 54

54 voltage passed Vcri. As illustrated in Figure 32, the saturation drain current was reduced ~15% for the Ni/Au gated HEMTs after the stress, with no obvious changes of the drain current for the Pt gated HEMTs. Similar trends were obtained for gate IV characteristics, as shown in Figure 33. There were permanent changes of both forward and reverse gate leakage characteristics for Ni/Au gated HEMTs. The Schottky barrier height was reduced from 1.17 to 0.53 V (Figure 33 A ) and the reverse gate leakage increased three orders (Figure 33 B ). By contrast, no changes in gate reverse and forward characteri stics were observed for HEMTs with Pt/Ti/Au gates. Besides the drain and gate IVs, the effects of the off state stress on the drain sub threshold characteristics were also investigated. The subthreshold leakage current, sub threshold slope and on/off drain current ratio are essential to the power added efficiency, linearity, noise figure and reliability of power amplifiers.4 Figure 34 shows the drain current as a function of the gate voltage for the unstressed A) Ni/Au gated and B) Pt/Ti/Au gate HEMTs biased at different drain voltage. At 10V drain voltage, the subthreshold leakage was ~mid 106 mA/mm for both devices. However, the subthreshold leakage current gradually increased as the drain voltage increased from 10 to 40V for the Ni/Au gated HEMTs. This was due to high reverse bias gate leakage of the Ni/Au HEMTs, as shown in Figure 33B There were minimal changes of the subthreshold leakage current levels for the Pt gated HEMTs, as shown in Figure 34B consistent with the low reverse bias gate leakage current as illustrated in Figure 33B Higher subthreshold leakage not only degrades the power added efficiency, linearity, noise figure of power amplifiers,4 it also significantly reduced the drain current onoff ratio of the HEMT. As shown in Table 31, the dr ain current onoff ratio of the Ptgate

PAGE 55

55 HEMTs stayed fairly constant around 1.5108 for the HEMTs biased at different drain voltages and this was due to low reverse bias gate leakage. By contrast, the drain current onoff ratio of the Ni/Au gate HEMTs decr eased from 1.16107 for the drain voltage of 10V to 6.29105 for the drain voltage of 40V. This was due to higher subthreshold drain leakage and reverse bias gate current. It was reported that the reverse bias gate current directly affected the subthreshol d slope.84 A subthreshold slope of around 65 mV/dec was achieved for the Pt/Ti/Au gated HEMTs, which was very close to the theoretical number of 60 mV/dec, as illustrated in Table 31. These results suggest that Pt/Ti/Au gate metallization maintains excellent reliability during the operating condition mentioned above. 3.4 Summary In conclusion, the significant improvement of AlGaN/GaN HEMT stability by using Pt based gate metallization instead of the conventional Ni/Au was demo nstrated. The off state critical voltage was increased from around 4565 to >100V, and changes of the drain current, drain current on/off ratio, Schottky barrier height and reverse bias gate leakage were minimized. The reverse bias gate leakage current and the stability between the gate metal contact and semiconductor contribute to the occurrence of a critical voltage, in addition to the electric field.

PAGE 56

56 Figure 31. Off state gate currents as a function of gate voltage for the HEM Ts fabricated with d i f f e r e n t gate metallization, A) Ni/Au or B) Pt/Ti/Au 0 -20 -40 -60 -80 -100 10-710-510-310-1101103 I G (A) I G (mA/mm) VGS (V)Ni-gated HEMTs VDS = +5V 10-910-710-510-310-1 0 -20 -40 -60 -80 -100 10-710-510-310-1101103 Pt-gated HEMTs VDS = +5V 10-910-710-510-310-1 I G (A) I G (mA/mm) VGS (V) (A ) ( B )

PAGE 57

57

PAGE 58

58 Figure 32. Drain IVs of HEMTs fabricated with d i f f e r e n t gate metallization measured before and after the off state stress. A) Ni/Au and B) Pt/Ti /Au 0 1 2 3 4 5 0 100 200 300 400 500 Ni-gated HEMTs Fresh StressedIDS (mA) I DS (mA/mm) VDS (V)VG = 0V, step = -1V0 20 40 60 80 100 0 1 2 3 4 5 0 100 200 300 400 500 Pt-gated HEMTs Fresh StressedIDS (mA) I DS (mA/mm) VDS (V)VG = 0V, step = -1V0 20 40 60 80 100 (A ) ( B )

PAGE 59

59 Figure 33. Schottky gate characteristics of the HEMTs fabricated with d i f f e r e n t gate metallization measured before and after the off state stress. A) Ni/Au or B) Pt/Ti/Au 0 1 2 3 10-910-710-510-310-1 I G (A) I G (mA/mm) V G (V) Pt-gated HEMTs Ni-gated HEMTs Fresh Stressed 10-1210-1110-1010-910-810-710-610-5 -50 -40 -30 -20 -10 0 10-910-710-510-310-1 IG (A) IG (mA/mm)VG (V) Pt-gated HEMTs Ni-gated HEMTsFresh Stressed10-1210-1110-1010-910-810-710-610-5 (A ) ( B )

PAGE 60

60 Figure 34. Subthreshold drain current of AlGaN/GaN HEMTs fabricated with d i f f e r e n t metallization biased at different drain voltage. A) Pt/Ti/Au and B) Ni/Au -6 -5 -4 -3 10-610-510-410-310-210-1100 VDS= 10V 20V 30V 40VIDS (A) IDS (mA/mm)VGS (V)Ni-gated HEMTs 10-910-810-710-610-510-4 -6 -5 -4 -3 10-610-510-410-310-210-1100 VDS= 10V 20V 30V 40V10-910-810-710-610-510-4 Pt-gated HEMTsIDS (A) IDS (mA/mm)VGS (V) (A ) ( B )

PAGE 61

61 Table 31. Summary of subthreshold slope and ON/OFF ratio of AlGaN/GaN HEMTs with Pt/Ti/Au and Ni/Au gate metallization before and after off state electric step stress VDS (V) Subthreshold slope (mV/dec) ON/OFF ratio Ideality factor Schottky barrier height (mV) Pt/Ti/Au 10 64.8 1.5610 8 1.61 1.25 40 65.3 1.4810 8 1.53 1.26 Ni/Au 10 78.8 1.1610 7 1.72 1.17 40 99.1 6.2910 5 4.62 0.53

PAGE 62

62 CHAPTER 4 INVESTIGATING THE EFFECT OF OFFSTATE STRESS ON TRAP DENSITIES IN ALGAN/GAN HIHG MOBIL ITY TRANSISTORS 4.1 Background The AlGaN/GaN heterostructure system has attracted a great deal of attention for highpower and high frequency applications .3 5 85 However, some undesirable issues induced by electron trapping at surface states of the semiconductor, such as current collapse86, 87 and gate leak age effects ,8890 have not been fully understood. Recently, intensive studies have been carried out to understand and eliminate the trapping effects in AlGaN/GaN high electron mobility transistors (HEMTs). Several different techniques have been utilized, including current mode deep level transient spectroscopy (DLTS),91, 92 conductance deep level transient spectroscopy (CDLTS) ,93, 94 transient drain current ,95 gate leakage current96 and threshold voltage97 measurements at different temperatures as well as frequency dispersion in capac itance and conductance analysis .49 To date, a number of different trap levels have been reported, such as 70 meV ,98 0.37 eV99 and 0.06 eV for donor like states related to N vacancies in AlGaN/GaN heterostruct ures, while holelike traps with activation energies of 0.4 and 0 .84 eV46 were also detected. The estimated trap density by the means of capacitancevoltage (C V) technique was 1.341012 /cm2eV at the pinchoff voltage ,100 and from bias dependent C V data, the values of Dit ranged from ~1011 to ~51012 /cm2eV .52 The density of trap states evaluated on the HFETs was 2.5101012 /cm2eV by frequency dependent capaci tance and conductance analysis .50 However, the investigation of trap states on both fresh and stressed AlGaN/GaN HEMTs has not been reported in detail. In this chapter, temperature dependent subthreshold slope measurements of AlGaN/GaN HEMTs in the temperature range from 300 to 573K have been performed.

PAGE 63

63 The increasing of subthreshold slope at elevated temperatures reveals two trap densities, which are attributed to traps located at the interface between the barrier layer and bulk layer or inside GaN bulk layer, as well as the increase in trap densities after electrical stressing. 4.2 Experimental The HEMT device structures were grown on semi insulating 6H SiC substrates and consisted of a thin AlN nucleation layer, 2.25 m of Fe doped GaN buffer, 15 nm of Al0.28Ga0.72N, and a 3 nm undoped GaN cap. Onwafer Hall measurements showed sheet carrier concentrations of 1.061013 cm2, mobility of 1907 cm2 /V s, and sheet d mesa isolation, Ti/Al/Ni/Au O hmic contacts alloyed at 850 finger Ni/Au gates patterned by lift off. The gate length was 1m, and gate width was 2150 m. Both source to gate gap and gat e to drain distances wer e 2 m. The devices exhibited typical maximum drain currents of 1.1 A/mm, extrinsic transconductance of 250 mS/mm at VDS of 10 V, threshold voltage of 3.6 V. An automated temperature control chuck from Wentworth was used to perform temperature dependent measurements. The base temperature was varied from room temperature to 300C and held constant during the measurement. The device DC characteristics were measured with a HP 4156 parameter analyzer. 4.3 Results and Discussion Figure 4 1 A shows the room temp erature drain and gate current currents as a function of the gate voltage for HEMTs prior to and after off state drain voltage stepstress at a constant gate bias voltage of 8 V. Figure 4 1 B shows the room temperature gate currents, IG, during the drain v oltage stepstress with a constant gate voltage of -

PAGE 64

64 8V. The stress conditions were set for holding 60 seconds at each drain voltage step, while grounding the source electrode and maintaining constant 8 V on the gate. The stress started at VDS = +5 V, and the voltage step was + 1 V. During the stepstress, gateto source leakage current, IGS, and gateto drain leakage current, IGD, were also measured. Between each stepstress, drain I V, extrinsic transconductance, gate forward current biased from 0 to 1.5 V and gate reverse current biased from 0 to 5 V, source and drain resistance were also recorded. The critical voltage, Vcri, of the off state step stress was defined as the onset of a sudden IG increase during the stress. During the off state stress, onc e the drain bias voltage reached Vcri, not only did the gate leakage current suddenly increase, as in Figure 4 1 B but also the saturation drain current dec reased, as previously reported.20 These permanent changes corresponded with pits formed along the edges of the gate electr ode and/or gate metal diffusion .66, 82 Table 4 1 shows the summary of subthreshold dr ain leakage current, the drain current onoff ratio and the slope of the subthreshold drain current before and after the off state stress at room temperature. Once the drain bias voltage reached Vcri, the drain current onoff ratio considerably decreased from 3.8 105 to 5.6 103. The drain current onoff ratio reduction degrades the charge modulation in the two dimensional electron gas channel as well as the power added efficiency, linearity, noise figure and reliability of power amplifiers. Both subt hreshold drain leakage current and the slope of sub threshold drain current were dominated by gate leakage current. As illustrated in Figure 4 1 A sub threshold drain leakage current and reverse bias gate leakage current displayed two order increases at VG = 4V after the HEMTs were stressed. The sub threshold drain leakage current was dominated by the reverse bias gate leakage

PAGE 65

65 current. As shown in Figure 4 1 B the magnitude of the threshold drain leakage current was equal to the reverse bias gate leakage c urrent. The slope of sub threshold drain current was reported to be highly dependent on the reverse bias gate leakage current .84, 101 The subthreshold slope almost doubled and increased from 98 to 158 mV/dec after the stress The drain current sub threshold slope, S, has also been used to quantify trap densities in the gate modulated region of metal oxide semiconductor field effect transistors (MOSFETs) and AlGaN/GaN HEMTs .101 The interface trap density can be extracted from the change of S with temperature. By analogy with Si MOSFETs and treating the AlGaN as the dielectric layer, the equations for trap density are given by: S T = k q ln ( 10 ) ( 1 + ) = C C (4 1) D = C q (4 2) where T is the temperature, k is Boltzmanns constant, q is the electron charge, ln is the natural logarithmi c symbol, is the ratio of capacitance associated with the interface traps and AlGaN layer capacitance, CAlGaN, and Dit is the interface trap density. The trap densities were extracted from the slope of S v.s. T plot, as shown in Figure 4 2. These temperature dependent behaviors of trap densities and gate leakage current in AlGaN/GaN heterostructures were attributed to surface hopping conduction through the traps .96 As illustrated in Figure 4 2, there was a transition temperature at 500K and two trap densities (Dit 1 and Dit 2) were identified for both reference and stressed devices in the temperature ranges of 300500K and 500 573K, respectively, based on drain current sub threshold slopes measured at different temperatures The estimated

PAGE 66

66 interface trap densities, Dit 1 and Dit 2, were 1.6 0.3 1012 and 8.1 0.3 1012 /cm2eV for the reference HEMTs, and 3.3 0.1 1012 and 9.2 0.5 1012 /cm2eV for the stressed devices, respectively The increases of trap densit ies for the stressed HEMTs as compared to the reference samples were due to additional traps generated during the stress providing paths for the gate leakage current, such as dislocation generation, gate metal diffusion, interface oxide intermixing with the semiconductor, and/or notch fo rmation around the gate edges. For the reference HEMTs, the increase of trap densities at higher temperature could be due to other deep traps being activated at the elevated temperatures, creating leakage current paths. We also noticed that the subthreshold drain leakage current and reverse bias gate leakage current for the reference HEMT measured at the temperatures above 493K were quite different from the ones measured at room temperature. Figure 4 3 shows the subthresh old drain leakage current and reverse bias gate leakage current of the reference HEMT measured at 300 and 573K. The sub threshold drain leakage currents were actually higher than the reverse bias gate leakage current for the reference HEMTs. Thus besides t he reverse bias gate leakage current, there were other leakage current paths involved in the increase of the subthreshold drain leakage currents, which would be the main cause for the increase of the subthreshold slope and the trap density level from Dit 1 to Dit 2. Ta ble 4 2 lists the summary of dc characteristics of the reference and stressed HEMTs measured at room temperature and 573K The saturation drain current, extrinsic transconductance, and drain current onoff ratio measured at 573K decreased for both reference and stressed HEMTs as compared to the room temperature results. However, there was also a noticeable increase in device

PAGE 67

67 isolation current for the HEMTs measured at 573K using device isolation testers. Since the source, drain and gate cont act pads of the HEMTs were connected through the active area of the device, the isolation current between the contact pads of the device could not be directly measured. An isolation tester with two 100 m 100 m Ohmic pads separated by 15 m isolated gap using ICP etching was used to examine the isolation current. Since the size of the contact pads was much smaller that the dimension of the source, drain and gate contact pads, the isolation current from the real device should be larger than that of the i solation tester s. As shown in Figure 4 4, the device isolation current increased proportional to the t emperature. Below the 493K, the isolation leakage was less than 0.5 A and the isolation leakage currents were lower than the gate leakage current. Thus, there was no impact observed on the subthreshold drain leakage. Once the measurement temperature was above 493K, the device isolation currents were comparabl e to the gate leakage current. For the reference HEMTs, at the higher measurement temperatures (>493K), the device isolation currents dominated the leakage currents of the device and the deep traps induced by the ion bombardment during the ICP device isolation etching were activated. Thus, the deep traps created during the mesa isolation etching were the main cause of the increase of the trap densities from Dit 1 to Dit 2 for the reference HEMTs. Proper thermal annealing after the ICP etching or employing ion implantationbased device isolation process should reduce the trap density Dit 2. For the stressed HEMTs, the reverse bias gate leakage current was only slightly lower than that of the subthreshold drain leakage current. Therefore, both traps created during the stress and those activated at the high temperature played a role in the increase of tr ap densities.

PAGE 68

68 The effect of drain bias voltage on the trap densities was also investigated. The devices were biased at drain bias voltages of 0.1, 2.5, 5 and 10V Figure 4 5 A shows the trap densities as a function of drain bias voltage. Both the Dit 1 an d Dit 2 increased with drain voltage, but Dit 1 was less se nsitive to drain bias voltage. At higher drain bias voltage, the maximum electric field on the gate edge close to the drain electrode increased, thus more hot electrons were generated and more traps creat ed. Figure 4 5 B shows the device isolation current as a function of the drain bias voltage and temperature. At higher bias voltage or higher temperature the isolation leakage current increased significantly. Therefore Dit 2 was more sensitive to the drain bias voltage. 4.4 Summary In conclusion, temperaturedependent subthreshold slope analyses were performed in order to investigate the trapping effects in AlGaN/GaN high electron mobility transistors (HEMTs). The sub threshold slope increased with temperature ranging from room temperature to 573K, there are two trap densities of 1.6 0.3 1012 and 8.1 0.3 1012 /cm2eV for the reference HEMTs and 3.3 0.1 1012 and 9.2 0.5 012 /cm2eV for the stressed HEMTs, respectively.

PAGE 69

69 Figure 41. DC characteristics and electrical stress data. A ) Sub threshold drain current and reverse bias gate leakage before and after the off state dr ain bias voltage step stress. B ) Gate currents as a function of stepdrain voltage during the off state stress. -6 -5 -4 -3 -2 10-410-2100102104106 I D I G (A) I D I G (mA/mm) VG (V) Reference Stressed ID IGVDS = +5V (A) 10-610-410-2100102 0 10 20 30 40 10-410-310-210-110010 1 (B) I G (A) I G (mA/mm) VDS (V) IG VG= -8V IGS IGD IGoff 10-710-610-510-410-3

PAGE 70

70 Figure 42. Slope of subthreshold drain current of the AlGaN/GaN HEMTs before and after off state drain bias voltage stepstress as a function of the ambient temperature. 300 350 400 450 500 550 600 0.1 0.2 0.3 0.4 Dit-2Dit-1 Reference StressedS (V/dec)Temperature (K)

PAGE 71

71 Figure 4 3. Sub threshold drain leakage current and reverse bias gate leakage current for reference HEMT measured at the temperatures of 300 and 573K. -6 -5 -4 -3 -2 10-410-310-210-1100101102103104 ID IG300 KID, IG (A) ID, IG (mA/mm)VG (V) VDS = +5V 573 K10-710-610-510-410-310-210-1100

PAGE 72

72 Figure 44. Device isolation current at a drain bias voltage of 10V and the gate leakage current at a gate voltage of 10V. 300 350 400 450 500 550 600 0 1 2 3 4 5 6 Sub-threshold leakage current Isolation leakage currentCurrent ( A)Temperature (K)

PAGE 73

73 Figure 45. The results of trap characteristics and device isolation current. A ) Trap densities of the HEMTs before and after the off state stress as a functi on of the drain bias v oltage. B ) Device isolation current as a function of drain bias voltage at different temperatures. 0 2 4 6 8 10 0 2 4 6 8 10 12 Dit-1 Reference StressedTrap Density ( 1012/cm2-eV )Drain Bias (V) Dit-2(A) 0 20 40 60 80 100 0 10 20 30 (B)Current ( A)Voltage (V)Start from 300K stop at 573K step = 20K T

PAGE 74

74 Table 4 1. Sub threshold drain leakage current, the slope of the sub threshold drain current and drain current on/off ratio of the HEMTs before and after the off state drain voltage stepstress Table 42. Room and higher Temperature drain saturation current maximum gm, drain current ON/OFF ratio, isolation current at 10V for both reference and stressed samples Sample Subthreshold drain leakage current (A) Subthreshold slope (mV/dec) ON/OFF ratio Reference 0.58 98 3.8 105 Stressed 37.2 158 5.6 10 3 Temperature Saturation current (mA) Maximum Gm (mS/mm) ON/OFF ratio Isolation current on isolation pads V=10V R. T. Ref. 220 72 3.8 105 2.6 nA Stressed 210 65 5.6 103 3.0 nA 573 K Ref. 110 42 2.5 104 Stressed 94 37 1.5 103

PAGE 75

75 CHAPTER 5 COMPARISON OF PASSIV ATION LAYERS FOR ALG AN/GAN HIGH MOBILITY TRANSISTORS 5 .1 Background Remarkable progress has been made in recent years in high performance AlGaN/GaN high electron mobility transistors (HEMTs) grown on a variety of substrates, including sapphire, SiC and Si .12, 84, 102113 Polarization effects in these structures lead to surface states that can have a significant detrimental impact on device performance. Unless ste ps are taken to passivate the device, positively charged surface donor states can trap negative charge and lead to the presence of a virtual gate that depletes the 2dimensional electron gas (2DEG) of carriers, reducing drain current and power performance. The most commonly used passivation layer is SiNX.45, 114, 115 Such layers are typically deposited by plasma enhanced chemical vapor deposition (PECVD) at temperatures near 300 C. The use of SiNX passivation typically restores 7080% of the dc current as measured by gate lag measurements Other dielectric passivation layers such as MgO or Sc2O3 deposited by plasmaassisted Molecular Beam Epitaxy (MBE) can provide even higher levels of current recovery under optimum conditions,116 but these are difficult t o integrate into the device processing sequence. The SiNX layer can also be used to define the T gate contact in a nitridefirst process which has both advantages and disadvantages at the device level. The overlap of gate metallization onto the surface of the SiNX can be employed to create a gate field plate and thereby reduce the peak electric field in the semiconductor, increasing the reverse breakdown voltage. However, the highk SiNX surrounding the gate also leads to parasitic loading and increases the fringing capacitance degrading the rf performance.

PAGE 76

76 Bilayer dielectric films consisting of thin SiNX and an overlayer of a lower dielectric constant material such as SiO2 have also shown good results .117 For aggressively scaled HEMTs with gate length <100nm, the aspect ratio of SiNX thickness to gate length may complicate accurate pattern transfer required for gate length definition when also used as the gate dielectric .118 For such devices, it is advantageous to make the dielectric very thin. For deposition of thin passivant films, atomic layer deposition (ALD) has received considerable recent attention. Many highk passivant films can be optimized as surface passivation and for use as gate insulators for high frequency GaN MOS HEMTs, leading to reduced gate leakage current relative to traditional Schottky gate HEMTs. In some cases, multi layer passivation structures have been effective (eg. Al2O3+SiNx),119122 although these add to the process complexity. Other approaches include AlN films as passivant / gate diel ectric and heat spreading layer .119, 120 In view of this past work, it is of interest to compare the passivation properties on AlGaN/GaN HEMTs of simple, single, thin layers of common oxides (HfO2 and Al2O3) deposited by ALD with conventional thick SiNx layers. We find the oxides are not as effective in passivating the surface states that cause cur rent collapse for this particular AlGaN/GaN epitaxial structure, but they do create less degradat ion of rf properties. 5 .2 Experimental The AlGaN/GaN HEMTs were fabricated on 6H SiC semi insulating substrate, with the following sequence of epitaxial layers : an AlN nucleation layer, a 1.8 m GaN buffer layer, a 1nm In0.10Ga0.90N backbarrier, a 15nm GaN channel and a top 22nm Al026Ga0.74N barrier. On wafer Hall measurements showed a sheet carrier concentration, sheet resistance, and mobility of 9.1 x 1012 cm2, 410 Ohms/square, and

PAGE 77

77 1670 cm2/V.s, respectively. An inductively coupled plasma mesa etch of ~1000 was performed to isolate adjacent devices. Ti/Al/Ni/Au ohmic metallization was annealed at 850C for 30 sec for source/drain contacts, producing a specif ic con tact resistance of 0.4 Ohm.mm. design with a gate length of 0.2um and width of 150m, source to gate and gateto drain distances of 1 m. Some control devices were left unpassivated, while others were passi vated with either 200 nm of SiNx deposited by a Plasma Therm 790 PECVD system or 7.5 nm of HfO2 or Al2O3 deposited by an atomic layer deposition system (Cambridge NanoTech, Inc, Fiji F200). The passivation thicknesses were chosen as being close to standard for the particular deposition process. Standard solvent precleans were done prior to passivation deposition. The passivation thicknesses and their dielectric constants had no measurable effect on the device parameters at these gate lengths. The dc charact eristics of the HEMTs were measured with a Tektronix curve tracer 370A and an HP 41 56 parameter analyzer The RF performance of the HEMTs was characterized with an HP 8723C network analyzer. Wafer scale maps of typical RF (fT) and dc (threshold voltage, Vth) performance are shown in Figure 5 1 for the unpassivated devices. 5 .3 Results and Discussion Figure 5 2 A shows the gate current voltage (I V) characteristics from the devices without passivation and those with the three different passivation layers in vestigated. The extracted ideality factors on the unpassivated devices were unphysically large (>5 when assuming thermionic emission as the d ominant conduction mechanism). Clearly there are multiple mechanisms present, such as tunneling and recombination. The passivated devices showed ideality factors of ~2 in all cases, indicative of recombination

PAGE 78

78 as the dominant transport mechanism in the heterojunctions. The SiNx passivated samples exhibited the highest reverse bias gate leakage current, although past reports have shown this to be highly dependent on the deposition conditi ons .123125 The use of ALD Al2O3 has been shown to reduce the gate leakage current .125 As shown in Figure 5 2 A both Al2O3 and HfO2 passivation layers effectively reduced the gate leakage current. The sub threshold leakage current for unpassivated and passivated devices is shown in the IDSVG plots of Figure 5 2 B All the passivation layers lead to a more negative threshold voltage, consistent with less depletion under the gate contact. Similarly, the transconductance and drain current of the passivated devices was higher, as shown in the transfer characteristics of Figure 5 2 C Typical IDSVDS characteristics are shown in Figure 5 3 for the unpassivated device ( a ) and for one passivated with SiNX ( b ). Fairly similar results were obtained with all the different passivation layers, namely, the drainsource current increased as was observed from the transfer characteristics and there were no kinks or other irregularities introduced that would suggest additional traps being introduced by the deposition process for these passivation films. Figure 5 4 summarizes the effect of th e passivation layers on threshold voltage, Vth, ( a ), unity current gain frequency, fT, ( b ) and maximum frequency of oscillation, fMAX, ( c). There is a larger negative shift in threshold voltage with SiNX which suggests it is more effective in r educing the surface depletion. However, the thin oxide films show the best improvement in RF characteristics due to their lower overall capacitance relative to the thick SiNX.

PAGE 79

79 We have employed gate lag measurements on the HEMTs as a metric for establishing the effecti venes s of the dielectric passivation.126 In this method, the drain current (IDS) response to a pulsed gatesource voltage (VG) is measured. Figur e 5 5 shows the normalized IDS as a function of drainsource voltage (VDS) for bo th dc and pulsed measurements. In this drain pulse data, VG was 0V, while the measurements used different f requencies and 10% duty cycle. Data is shown for the unpassivated HE MT (a ), for HfO2 ( b ), and SiNX ( c) passivants. The large differences between dc and pulsed drain currents for the unpassivated HEMT are consistent with the presence of surface traps that deplete the channel in the access regions between the gate and drain contacts. After nitride or oxide deposition, the HEMTs showed an increase in drainsource current in the dc mode which is consistent with passivation of surface states. The percentage reductions in drain current at 3V drainsource voltage are shown in Tabl e 5 1 The SiNX is the most effective at minimizing the trapping evident from the drain pulse measurements and is due to an increase in positive charge at the passivation layer/AlGaN interface, resulting in an increase in effective sheet carrier density in the channel. Figure 5 6 shows the corresponding gatelag measurements, with a marked improvement in drain current response for the passivated devices relati ve to the unpassivated device. This is clear evidence for the assumption that surface states are th e cause of the gatelag phenomena and also that passivation layers mitigate this problem. Once again, the SiNX is the most effective at reducing the effect of the surface traps.

PAGE 80

80 We also used double pulse measurements, in which VDS was pulsed from 10V to 5V while VG was simultaneously pulsed from 6V to 1 V in steps of 1V, at 10% duty cycle .127, 128 Usually, the drain current at gate bias voltages close to the threshold voltage would not decrease during the gate pulse measur ement, as shown in Figure 5 6 This is due to an insufficient population of hot electrons being generated at these conditions (5 V of drain bias voltage and gate voltage pulsed from the pinch off voltage to 1 1.5 V above the t hreshold voltage). H owever, during the double pulse m easurement, the hot electrons can be generated by the electrons in the gate leakage current being accelerated by the higher field present between the 10 V of drain bias voltage and 6 V of gate voltage during the off state. Thus, eno ugh hot electrons are injected into the surface between the gate and drain electrode and create a virtual gate in this region, suppressing the drain current. The resulting data is shown in Figure 5 7 for unpassivated HEMTs ( a ) and devices with HfO2 ( b ) or SiNX ( c ) passivation. These pulsing conditions led to there being no current in the unpassivated device, while those with surface passivation showed various degrees of recovery. The double pulse measurement is a valuable method for examining the effectiveness of the device passivation and SiNX was once again the best choice under dc conditions. 5.4 Summary Alternatives to the usual thick SiNX passivation films on AlGaN/GaN HEMTs were examined. Both HfO2 and Al2O3 thin films deposited by ALD produce improvem ents in drainsource current due to a reduction of surface depletion effects but are less effective than SiNX for the given AlGaN/GaN epi structure. However, the oxides produce superior RF performance to SiNX and should offer advantages with respect to gat e definition because of their reduced aspect ratios.

PAGE 81

81 Figure 51. Wafer scale maps for unpassivated HEMTs. A) fT and B) Vth. Ft@Gmp (GHz) 5 10 15 5 10 15 152(7<<137>>8) 90.1%[0] 0 10 20 30 22.74 25.96 29.17 32.39 35.6 38.82 42.03 45.25 (A) Vth (V) 5 10 15 5 10 15 152(13<<134>>5) 88.2%[0] 0 10 20 30 -3.688 -3.559 -3.431 -3.303 -3.174 -3.046 -2.918 -2.79 (B )

PAGE 82

82 Figure 5 2. DC characteristics of unpas sivated and passivated HEMTs. A) Gate I V, B) IDSVG and C) transfer characteristics -10 -8 -6 -4 -2 0 2 10-710-610-510-410-310-210-1 No passivation Al2O3 (7.5nm) HfO2 (7.5nm) Si3N4 (200nm) I G (A) I G (mA/mm) V G (V)10-1110-1010-910-810-710-610-5 (A ) -6 -5 -4 -3 -2 -1 0 1 10-610-410-2100102104 IDS (A)VG (V) No passivation Al2O3 (7.5nm) HfO2 (7.5nm) Si3N4 (200nm)VDS = +5VIDS (mA/mm)10-1010-910-810-710-610-510-410-310-210-1100 (B ) -6 -5 -4 -3 -2 -1 0 1 0 200 400 600 800 1000 V G (V) VDS = +5V IDS Gm No passivation Al2O3 (7.5nm) HfO2 (7.5nm) Si3N4 (200nm)G m (mS/mm) I DS (mA/mm) 0 200 400 600 (C )

PAGE 83

83 Figure 5 3. IDSVDS characteristics o f u n p a ssi v a t e d a n d p a ssi v a t e d d e v i ce s. A) the unpassivated device and B) for one passivated with SiNX. 0 1 2 3 4 5 0 200 400 600 800 1000 1200 IDS (mA) IDS (mA/mm)VDS (V)VG=+1V, -1V step(A)0 20 40 60 80 100 120 0 1 2 3 4 5 0 200 400 600 800 1000 1200 (B)IDS (mA) IDS (mA/mm)VDS (V)VG=+1V, -1V step 0 20 40 60 80 100 120

PAGE 84

84 Figure 54. DC and RF characteristics of HEMTs with different passivation layers. A) Threshold voltage, B) unity current gain frequency, fT and C) maximum frequency of oscillation, fMAX, as a function of type of passivation film. -2.5 -3.0 -3.5 -4.0 -4.5 Si3N4HfO2Al2O3VT (V)None (A ) 20 40 60 80 Si3N4HfO2Al2O3FT@GmP (GHz)None (B ) 60 80 100 120 140 160 Si3N4HfO2Al2O3Fmax@GmP (GHz)None (C )

PAGE 85

85 Figure 5 5. Drain pulse measurements on u n p a ssi v a t e d a n d passivated AlGaN/GaN HEMTs A) unpassivated, B) HfO2 passivated and C) SiNX passivated AlGaN/GaN HEMTs. VG is 0V at a 10% duty cycle. 0 1 2 3 4 5 0 20 40 60 80 100 0 20 40 60 80 100 IDS at VG=0V 100Hz 10kHz 500kHzNormalized IDS (%)VDS (V)Si3N4 (C) 0 1 2 3 4 5 0 20 40 60 80 100 0 20 40 60 80 100 IDS at VG=0V 100Hz 10kHz 500kHzNormalized IDS (%)VDS (V)HfO2 (B) 0 1 2 3 4 50 20 40 60 80 100 0 20 40 60 80 100 IDS at VG=0V 100Hz 10kHz 500kHzNormalized IDS (%)VDS (V)No Passivation (A)

PAGE 86

86 Figure 56. Gate pulse measurements on u n p a ssi v a t e d a n d passivated AlGaN/GaN HEMTs A) unpassivated B) HfO2 passivated and C) SiNX passivated AlGaN/GaN HEMTs. VG is switched from 5V to the value shown on the X axis at a 10% duty cycle. -6 -5 -4 -3 -2 -1 0 1 0 20 40 60 80 100 IDS (DC result) 100Hz 10kHz 100kHz VDS = +5VNormalized IDS (%)VG (V)No Passivation (A) -6 -5 -4 -3 -2 -1 0 1 0 20 40 60 80 100 HfO2 IDS (DC result) 100Hz 10kHz 100kHz VDS = +5VNormalized IDS (%)VG (V) (B) -6 -5 -4 -3 -2 -1 0 1 0 20 40 60 80 100 Si3N4 IDS (DC result) 100Hz 10kHz 100kHz VDS = +5VNormalized IDS (%)VG (V) (C)

PAGE 87

87

PAGE 88

88 Figure 5 7. Double pulse measure ments with drain pulsed from 10V to 5V while simultaneously pulsing the gate from 6V to the value shown on the x axis. The plots are for A) unpassivated, B) HfO2 passivated and C) SiNX passivated AlGaN/GaN HEMTs. -6 -5 -4 -3 -2 -1 0 1 0 20 40 60 80 100 IDS (DC result) 100Hz 10kHz VDS = +5VNormalized IDS (%)VG (V)No Passivation (A) -6 -5 -4 -3 -2 -1 0 1 0 20 40 60 80 100 Si3N4 IDS (DC result) 100Hz 10kHz VDS = +5VNormalized IDS (%)VG (V) (C) -6 -5 -4 -3 -2 -1 0 1 0 20 40 60 80 100 HfO2 IDS (DC result) 100Hz 10kHz VDS = +5VNormalized IDS (%)VG (V) (B)

PAGE 89

89 Table 51. Summary of dc characteristic s of unpassivated and passivated HEMTs. The percentage current losses were calculated from the 500 kHz data. Passivation Layer Dielectric Constant Ideality Factor % current loss drain pulse % current loss gate pulse % current loss double pulse none 1 ~5 1 8 36 100 Al2O3 ~9 ~2 10 26 57 HfO2 2025 ~2 5 23 57 SiNX 6 10, 7.5 for ideal value ~2 0 21 21

PAGE 90

90 CHAPTER 6 EFFECT OF BUFFER DESIGN ON THE RELIABILITY OF ALGAN/GAN HIGH ELECTRON MOBILITY TRANSISTORS 6.1 Background There are increasing needs for high power, high efficiency and linearity amplifiers for communication systems and radars for defense applications. Gallium Nitride (GaN) based High Electron Mobility Transistors (HEMTs) are the basis for the next generation of power transistor technology and hav e exhibited impressive attributes, such as high mobility, high saturation velocity and power density for high voltage and high switching frequency applications.7 8 10, 129 As a result of the large conduction band discontinuity between Al0.25GaN (~4.2 eV) and G aN (3.4 eV), both high electron mobility and high electron saturation velocity have been demonstrated in the AlGaN/GaN material system, with values of over 1500 cm2/V s and 2.5107 cm/s, respectively. These parameters are beneficial for high frequency operation. The presence of both a strong piezoelectric effect and spontaneous polarization in III nitrides lead to a high carrier concentration (> 1013 cm2), and in consequence a high current density can be achieved in AlGaN/GaN heterostructures without conv entional doping.14 130 Moreover, GaN has a large band gap and consequent high breakdown electric field (~3 MV/cm), which make it cap able of handling high voltages. These merits of high current density and high breakdown voltage enable this material to be an excellent candidate for high power applications. Silicon carbide (SiC), sapphire and freestanding GaN are the most widely used sub strates for GaN HEMTs epitaxy. SiC is a good candidate due to its h igh thermal conductivity. Sapphire substrates have lower cost, but their low thermal conductivity

PAGE 91

91 limits their applications for power transistors. Freestanding GaN technology is still evolving. During heteroepitaxy, impurities are introduced at the growth interface and may cause leakage paths, which degrades on/off ratio and off state breakdown voltage.131, 132 Several solutions have been developed to reduce buffer leakage, such as deliberate introduction of deeplevel impurities, Fe or C, into the GaN buffer133, 134 and also different epitaxial structure designs.135137 The dislocation densities due to lattice mismatch between GaN and commonly used SiC or sapphire substrates are on the order of 108 109 cm2, and this generally reduces with GaN buffer thickness. As reported by Hinoki et al.138, 139 the off state b reakdown voltage increased and leakage current decreased with increasing thic kness of the GaN buffer layer. It was reported that higher densities of edgetype dislocations contributed to the improvement of breakdown voltage.136 It was reported t hat UV induced electron detrapping increased the peak electric field around gate edges, resulting in a lower critical voltage, and also suggested that an increased number of traps might improve HEMTs reliability.140 Ivo and coworkers also studied the effect of buffer designs on the reliability of AlGaN/GaN HEMTs by electroluminescence (EL) and electrical measurements, which indicated that higher robustness could be achieved when Al0.05GaN barrier layers were used.141 In this chapter, the reliabi lity of AlGaN/GaN HEMTs fabricated on SiC substrates composite AlGaN/GaN buffers were used in this work. The effect of various buffer structures on device reliability, iso lation breakdown voltage and off state drain breakdown voltage are reported.

PAGE 92

92 6.2 Experimental AlGaN/GaN HEMTs were grown by Metal Organic Chemical Vapor Deposition (MOCVD) on SiC substrates. The buffer structure was varied while the active layers of the HE For composite buffer structure, the growth of AlGaN was followed by GaN. This was followed by a 55nm thick GaN channel, 21nm of A l0.25Ga0.75N and finally a 2.5nm GaN cap, as shown in Figure 6 1 The nominal growth temperature for the GaN buffer and Al GaN barrier layers was 1030C. Transistor fabrication began with deposition and patterning of Ti/Al/Ni/Au Ohmic metallization and subs equent RTA in flowing N2 at 850C. The contact resistance, specific contact resistivity, and sheet resistance were extracted from tr ansmission line measurements (TLM) using 100 100 2 pads separated by 5, 10, Device isolation was accomplished by multiple energy N+ implantation. Schottky contacts were formed by deposition of 1000 Ni/Au gates. DC curr ent voltage (I V) characteristics were recorded using a HP4156C parameter analyzer. 6.3 Results and Discussion Figure 62 shows the typical transfer characteristics of HEMTs fabricated on different buffer layers, measured at VDS = +5V. The drain current at VG = 1.5V was buffer layer was more negative than those with the other buffer designs, while the transconductance peak of the former was higher than others. As illustrated in Figure 6 3, the drain and gate currents as a function of gate voltage for HEMTs fabricated on

PAGE 93

93 different buffer layers were measured at VDS = +5V. Sub threshold drain leakage current was dominated by reverse gate leakage current with the channel off. Both subthreshold drain leakage current and reverse gate leakage current were very similar for all HEMTs with the various buffer designs. Drain current voltage (I V) characteristics of AlGaN/GaN HEMTs fabricated on different types of structures are shown in Figure 6 4, which were recorded with VDS from 0 to 5V and VG starting from +1V with a step of 1V. The saturation drain current of HEMTs with thick GaN and composite buffers were around 5Figure 6 5 shows gate I V characteristics of HEMTs fabricated on different buffer structures. No apparent difference of gate I Vs was observed, the extracted Schottky barrier height (SBH) and ideality factor of HEMTs with thick GaN and composite buffer structures were 1.28 V and 1.85, while they were 1.36 V and 1.61 with the thin GaN buffer layer. The effective barrier heights and ideality were obtained from the relat ionship as shown in E quation 61. J = A T exp ( ) eV n K T 1 (6 1) W here A**, the Richardson constant, is 29.2 A/cm2K2 for n GaN, J the current density, T the measured temperature, B the barrier height, n the ideality factor and KB is Boltzmanns constant. Off stat e drain electrical stepstress was employed to evaluate the reliability of HEMTs fabricated on the various buffer layers.21 Figure 6 6 shows the gate current dur ing electrical stepstressing. The gate current, IG, was plotted as a function of stressed drain v oltage. Each drain voltage step was constant for 60 seconds, while grounding the source electrode and fixing gate voltage at 6V. The stress started at 5 V

PAGE 94

94 of drain voltage and the voltage step was kept at 1 V. During the stepstress, besides monitoring IG, gate to source leakage current, IGS, and gateto drain leakage current, IGD, were also measured. Due to the low drainto source current during off state, self heating effects were negligible and had no effect on device performance. In the early stage of electrical stress, all IG curves almost overlapped, since they were dominated by the reverse gate current (Figure 6 5 ). The critical voltage, Vcri, of the off state step stress was defined as the onset of a sudden IG increase during the stress. As shown in Figure 6 6, once reaching the critical voltage, besides the increase of gate current, drain saturation current was also decreased significantly. This was due to permanent damages form ed around the gate electrode. Cracks appeared on both the source and dra in sides of the gate edges.82 Diffusions of nickel gate contact and native oxygen layer at metal/AlGaN interface occurred and threading dislocations provided additional pathways for the increased gate leakage current.142 The Vcri for HEMTs fabri cated on 2 50V, 70 90V and 6580V, respectively. The devices with thicker GaN buffer layers sh owed a lower critical voltage. Between each step stress, saturation drain I Vs of the HEMTs with different buffer layer were r ecorded, as shown in Figure 6 7. Since the unpassivated HEMTs were intentionally employed for this study to compare the effect of buffer layer structures on the device performance, HEMTs with all these three buffer layer structures exh ibited serious drain current collapse issue. However, the HEMTs with 2 m thick GaN buffer only showed around 20% degradation even after the HEMTs been stressed at 40V a pplied to the drain electrode. HEMTs with 1 m thick GaN and GaN/AlGaN composite buffer layer structures showed 50% drain current reductions.

PAGE 95

95 Besides off state drain electrical stepstress, off state drain breakdown voltage (Voff state) and isolation breakdown voltage (Viso) were characterized as well. The off state drain breakdown voltage is shown in Figure 6 8 A which was measured far below the typical threshold voltage of VG = 8V. Besides the buffer layer, field plate also played an important role on critical voltage of the off state step stress as well as off state drain breakdown voltage. Field plate altered the electrical field distribution around the gate and lowered the electrical field around the gate edges; thus improved both critical voltage of the off state step stress and off state drain breakdown voltage. Isolation breakdown voaps, as shown in Figure 68 B To protect devices from breakdown, compliance currents of state breakdown voltage and isolation breakdown volta ge measurements, respectively. Just as the electrical stress results, the devices with thicker GaN buffer layers demonstrated the lowest Voff state values. The Voff state for HEMTs with thin GaN and composite buffers were ~100V, however, this value was 5060V for those devic es with thick GaN buffers. A similar trend was observed in the isolation breakdown voltage measurements, with the highest Viso achieved based on thin GaN or composite buffer designs with a value of 600700V, while a much smaller Viso of ~200V was measured on HEMTs with the thick GaN buffer layers. The critical voltage for electrical degradation is strongly dependent on electric field.142 During off state electrical stress, the peak electric field was located at both sides of the gate edges. Typically, the peak field at the drain side of gate edge was higher than that at source side, since the drain w as biased while the source was gr ounded in device applications. Additional mechanical stress will be induced due to the presence of

PAGE 96

96 this high electrical field in the piezoelectric GaN and AlGaN layers.22, 39 When the magnitude of the electric field reached threshold levels, a variety of degradation mechanisms such as cracks, notc hes, gate metal diffusion and corresponding threading dislocations could be created.40, 82 Finally, irreversible degradation in dc characteristics will be observed when this degradation progresses. When the thickness of the buffer layer is reduced, the defect density will be increased.138 139 Those defects act as trap centers, which remove free electrons from the channel and cause the formation of an additional virtual gate,45 especially, in the acces s regi on between the gate and drain. Consequently, the presence of a higher density of defects further extends the depletion region into the buffer layer, and the peak electric field at drain side of gate edge is reduced. The peak electric field is therefo re reduced for HEMTs fabricated with thin buffer layers as compared to those with thick buffer layer at a given drain bias and this leads to a higher critical voltage for the onset of leakage current and better apparent reliability during bias stressing. AlGaN plays as a transition layer between SiC substrate and GaN buffer layer since the lattice constant of AlGaN is more close to SiC as compared to GaN. By inserting the AlGaN between the substrate and GaN, the dislocation density in the GaN can be reduced. Although the thin and composite buff er structures have the same thickness, the density in the GaN layer is less for the latter which was also confirmed by pulse measurement in the following paragraph. Therefore, a higher critical voltage was demonstrated by HEMTs fabricated with thin GaN buffer layer. Gate pulse measurements were also employed to evaluate the material quality of the HEMT with different buffer layer structures. In this study the drain current (IDS)

PAGE 97

97 response to a pulsed gatesource voltage (VGS) was measured. The normalized IDS as a function of VGS in both pulsed and dc mode for HEMTs is shown in Figure 6 9 During the pulse measurement, the VGS was pulsed from 5V to the values shown on x ax is at different frequencies of 100 and 10 KHz with 10% duty cycle, while drain was kept constant at +5V. The reduction of the drain current in the pulsed mode as compared to the drain current in the dc mode was due to the presence of surface traps in the access region between gate and drain contact. Usually, the traps have some specific time constants and could not respond above certain frequency. Thus the pulsed drain current would be lower than the dc drain current. As shown in Figure 6 9, HEMTs with 1 m GaN buffer layer exhibited largest reduction of the drain current during the pulse measurement indicating more traps in this structure. Figure 6 10 illustrates the room temperature photoluminescence (PL) spectra of HMETs with different buffer layers. All sample exhibited a band edge (BE) emission at around 3.4 eV. A broad yellow luminescence (YL) band associated with deep levels including carbon impurity,143 Ga vacancy144 Ga interstitial145 and N antisite146 at around 2.2 eV (560 nm) was n ot observed for these samples. This implied that the qualities of the HEMT structures with these three buffer layers were reasonably good. The shoulder of the main BE peak ranging from range 3.25 to 3.35eV was originated from N vacancy.147 Besides BE emission, there were very low intensity broad blue luminescence (BL) b a nds centered at 2.852.95 eV. The insert PL spectra illustrated in Figure 6 10 are the enlarged spectra of these BL bands. These bands were often observed in the PL of GaN grown by MOCVD and were attributed to the transitions from the conduction band or a shallow donor.148 Among these three buffer layer, BL band

PAGE 98

98 showed a lowest intensity for the HEMT structure with 2 m GaN buffer and these results were consistent with HEMT dc characteristics and pulse measurements. 6.4 Summary T he effect of buffer designs on the reliability of AlGaN/GaN HEMTs by off state electrical drain stepstress was investigated. Three types of buffer layers were used in osite AlGaN/GaN buffer layer. The devices fabricated on thicker GaN layer s showed the lowest Vcri of 30 50V, while Vcri 65 80V was achieved with composite buffers and 7090V for those with thin GaN buffers. Similar trends were observed in off state breakdown voltage and isolation breakdown voltage measurements. The depletion r egion in HEMTs with thinner GaN buffers was extended into the buffer layer due to the presence of a higher density of defects, which was responsible for the improvement of device reliability.

PAGE 99

99 Figure 61. Schematic of epitaxial structures of Al GaN/GaN HEMTs

PAGE 100

100 Figure 62. Transfer characteristics of HEMTs fabricated on different buffer layers. -5 -4 -3 -2 -1 0 1 0 200 400 600 1 m GaN 2 m GaN AlGaN/GaN VDS = +5V G m (mS/mm) I DS (mA/mm) VG (V) 0 50 100 150 200

PAGE 101

101 Figure 63. Subthreshold drain I V characteristics of HEMTs fabricated on different buffer layers. -5 -4 -3 -2 -1 0 1 10-810-610-410-2100102104106 I DS I G (A) 1 um GaN 2 um GaN AlGaN/GaN VDS = +5VI DS I G (mA/mm) VG (V)10-1010-810-610-410-2100

PAGE 102

102 Figure 64. Drain characteristics of HEMTs fabricated on different buffer layers. 0 2 4 6 8 10 0 200 400 600 800 1 um GaN 2 um GaN AlGaN/GaNIDS (mA) I DS (mA/mm) V DS (V) VG = 1V, in -1 step 0 40 80 120 160

PAGE 103

103 Figure 65. Gate characteristics of HEMTs fabricated on different buffer layers. -10 -8 -6 -4 -2 0 2 10-810-710-610-510-410-310-210-1 1 um GaN 2 um GaN AlGaN/GaNI G (A) I G (mA/mm) VG (V)10-1110-1010-910-810-710-610-5

PAGE 104

104 Figure 66. Off state drain stepstress results of HEMTs fabricated on different buffer layers, measurements were conducted at VG = 6V. 0 20 40 60 80 100 10-610-510-410-310-210-1100101 VG = -6V 1 um GaN 2 um GaN AlGaN/GaN I G (A) I G (mA/mm) VDS (V)10-910-810-710-610-510-410-3

PAGE 105

105 Figure 67. Some of the saturation drain I Vs recorded between each step stress for HEMTs fabricated on different buffer layers. 0 1 2 3 4 5 0 20 40 60 80 100 120 Normalized IDS (%)VDS (V) Fresh 5V 10V 20V 30V 40VVG = 0V 1 m GaN buffer(A) 0 1 2 3 4 5 0 20 40 60 80 100 120 Normalized IDS (%)VDS (V) Fresh 5V 10V 20V 30V 40VVG = 0V 2 m GaN buffer(B) 0 1 2 3 4 5 0 20 40 60 80 100 120 Normalized IDS (%)VDS (V) Fresh 5V 10V 20V 30V 40VVG = 0V AlGaN/GaN buffer(C)

PAGE 106

106 Figure 68. The results of off state breakdown voltage and isolation breakdown voltage measurements of HEMTs fabricated on different buffer layers. A) off state breakdown voltage and B) isolation breakdown 0 50 100 150 10-210-1100101102103104 (A) 1 m GaN 2 m GaN AlGaN/GaNVoltage (V) Current ( A) 0 200 400 600 800 1000 10-210-1100101 (B)Current ( A)Voltage (V) 1 m GaN 2 m GaN AlGaN/GaN On isolation pads with 5 m gap

PAGE 107

107 Figure 69. Normalized drain current, IDS, as a function of VGS for both pulsed and dc modes. -5 -4 -3 -2 -1 0 1 0 20 40 60 80 100 0 20 40 60 80 100 DC result 100 Hz 10 KHzVDS = +5VNormalized IDS (%)VG (V)AlGaN/GaN buffer(C) -5 -4 -3 -2 -1 0 1 0 20 40 60 80 100 0 20 40 60 80 100 DC result 100 Hz 10 KHzVDS = +5VNormalized IDS (%)VG (V)2 m GaN buffer(B)

PAGE 108

108 Figure 610. Room temperature photoluminescence (PL) spectra of HMETs with different buffer layers. Insert: Enlarged blue luminescence (BL) bands of those HEMTs with different buffer layers.

PAGE 109

109 CHAPTER 7 EF FECTS OF PROTON IRRA DIATION ON GAN BASED HIGH ELECTRON MOBILITY TRANSISTORS 7.1 The Effects of Proton Irradiation on the Reliability of InAlN/GaN High Electron Mobility Transistors 7.1.1 Background InAlN/GaN high electron mobility transistors (HEMTs) appear to be an excellent candidate to replace conventional AlGaN/GaN heterostructures in some electronics applications. Promising DC, RF and output power performances of InAlN/GaN HEMTs on Si, sapphire and Si C substrates have been reported,149160 which make them suitable for high power and high frequency applications such as broadband communication and power flow control. InAlN with an In mole fraction of 0.17 can be grown lattice matched to GaN, which eliminates the strain present in the AlGaN/GaN heterostructure system and this should be ben eficial for device reliability .21 Due to the existence of large spontaneous polarization between InAlN and GaN, a high density two dimensional electron gas (2DEG), above 2.5 1013 cm2 can be achieved,161 leading to higher current densities and higher powers compared to typical AlGaN based HEMTs. Recently, a record current density of 2.5 A/mm at VG = +2V was reported with 6.9 nm barrier thickness and gate l ength (LG) of 100 nm .160 In addition, a thin barrier layer assists in reducing short channel effects in high frequency applications .162 For space applications, devices are always exposed to harsh conditions, including high energy proton, gamma ray and X ray fluxes. Thus far, the effect of proton irradiation on InAlN/GaN device performance has been investigated by several groups. Lo et al reported the degradation of DC performance of InAlN/GaN HEMTs after 5 MeV proton irradiation with doses varying from 21011 to 21015 cm.163 Kim et al. studied

PAGE 110

110 the influence of proton irradiation on InAlN/GaN HEMTs grown on SiC substrates, which were subjected to 5 to 15 MeV high energy protons with a fixed 51015 cm2 fluence.164 Irradiation at lower energy was found to degrade the direct current (DC) current voltage (I V) characteristics more severely than higher energy irradiation, because more defects were formed in the vicinity of the 2dimensional electron gas when lower energy was applied. The main degradation mechanism by proton irradiati on is displacement damage. Defect centers are introduced during the collisions between incident protons and nuclei of the lattice atoms. These defect centers have significant capture cross sections for free carriers, resulting in the reduction of carrier density and c onductivity of irradiated HEMTs .165167 The mobility is strongly affected by interface roughness168, 169 and the scattering from defect centers in the vic inity of the channel can also degrade mobility .166, 170 Analyses of degradation of lattice matched InAlN/GaN under different conditions have been reported.171 172 InAlN/AlN/GaN heterostructure field effect transistors were stressed under high electric field at room temperature. The degradation was attributed to the buildup of hot phonons, which caused loca l heating and defect generation.171 A comprehensive study, including off state stress, semi on stress and negative gate bias stress were performed by Kuzmik et al. Irreversible damage was found for the off state biasing and for the semi on stresses when draingate voltage was over 38V. The damage was considered as being due to hot electrons, which were injected into the GaN buffer layer underneath the gate and either cr eated defects or ionized existing defect states .172 There have also been reports of improvement of breakdown voltage in

PAGE 111

111 AlGaNbased HEMTs by proton irradiation.173 However, there is no report on the reliability of pro ton irradiated InAlN/GaN HEMTs. In this chapter the drain and gate I V characteristics of reference and protonirradiated InAlN/GaN HEMTs as well as the resultant performance of the HEMTs after off state drainvoltage stepstress biasing cycles were reported 7.1.2 Experimental The HEM T structures were grown with a Metal Organic Chemical Vapor Deposition (MOCVD) system, starting with a thin AlGaN nucleation layer, followed with a 1.9 m low defect carbondoped GaN buffer layer, 50 nm undoped GaN layer, 10.2 nm undoped InAlN layer with a 17% of In mole fraction, and capped with a 2.5 nm undoped GaN layer. The samples were all grown on three inch diameter, c plane sapphire substrates. Hall measurements on the as grown structures showed sheet carrier densities of 2.1 1013 cm2 and the cor responding electron mobility of 1000 cm2/ V s. Device fabrication began with the Ohmic contact deposition with the standard lift off e beam evaporated Ti/Al/Ni/Au based metallization, and the samples were subsequently annealed at 800C for 30 s under a N2 ambient. A typical contact resistance of 0.6 mm was obtained using the transmission line method (TLM). Multiple energy and dose nitrogen implantation was used for the device isolation and photoresist AZ1045 was used as the mask to define the active regio n of the devices. Isolation currents were less than 10 nA at 40 V of bias voltage across two 100 m 100 m square Ohmic contact pads separated by a 5 m implanted gap. 1 m gates were defined by lift off of e beam deposited Pt/Ti/Au metallization. Ti/Au metallization was utilized for the interconnect metals for source, gate, and drain electrodes. The

PAGE 112

112 transistors were passivated using 400 nm of the plasmaenhanced chemical vapor deposited (PECVD) SiNx at 300C, followed by opening of contact windows using fluorine based plasma etching. The DC characteristics of the HEMTs were measured with a Tektronix curve tracer 370A and an HP 4156 parameter analyzer. All the samples were proton irradiated in a vacuum chamber at room temperature with the MC 50 Cyclotron at the Korea Institute of Radiological and Medical Sciences. Proton beam energy was controlled from 15 to 5 MeV by inserting an aluminum degrader. The samples were mounted with carbon tape, where the front face aimed at the proton beam, which means that growth direction of the samples is parallel to the direction o f the proton beam. The dc characteristics of the HEMTs were measured w ith HP 4156 parameter analyzer. 7.1.3 Results and Discussion Fig ure 7 1 shows the gate current during typical off state step stresses of InAlN/GaN HEMTs prior to and post proton irradiation. The gate current, IG, was plotted as a function of stressed drain voltage. The devices were stressed for 60 seconds at each drain voltage step, while grounding the source electrode and a constant voltage of 6V applied to the gate electrode. The stress started at 5 V of drain voltage and the voltage step was kept at 1 V. During the step stress, besides monitoring IG, gate to source leakage current, IGS, and gateto drain leakage current, IGD, were also measured. Between each step stress, drain I V, extrinsic transconductance, gate forward current biased from 0 to 1.5 V and gate reverse current biased from 0 to 10 V, were recorded. Self heating effects were negligible based on the low drainso urce currents under our test conditions. The critical voltage, Vcri, of the off state step stress was defined as the onset of a sudden IG increase during the stress. Typical Vcri for electrical degradation of

PAGE 113

113 virgin (unirradiated) HEMTs ranged from 45 to 55 V. By sharp contrast, no such critical voltage was detected for devices after proton irradiation even the drain was biased to +100 V, which is the limit of our apparatus. The same results were observed for devices post 10 and 15 MeV proton irradiation as well as the HEMTs exposed with different doses ranging from 2 1011 to 2 1015 cm2 of protons at fixed energy of 5 MeV. During the off state stress, before the gate bias voltage reached the critical voltage, there was no degradation observed for both gate and drain I Vs. Once the drain bias voltage reached Vcri, not only did the gate reverse bias leakage current suddenly increase, as illustrated in Figure 7 1, but also the saturation drain current decreased, as previously reported in AlGaN/GaN21 and InAl N/GaN structures .172 As shown in Figure 72, the saturation drain current was reduced ~12% for the unirradiated HEMTs after the stress. There were no obvious changes of t he drain current for irradiated HEMTs, as illustrated in Fig ure 7 3. Besides drain I V characteristics, the gate I V characteristics exhibited a similar trend, as shown in Figure 74. Although the gate current of the irradiated HEMT was much higher as com pared to the unirradiated HEMTs, there were no changes in gate reverse and forward characterist ics after the off state stress. The decrease of drain saturation current and increase of reverse bias gate leakage current of irradiated devices were attributed to the reduction of sheet carrier concentration and carrier saturation velocity caused by the defects generated during th e proton implantation .163167 On the co ntrary, the unirradiated HEMTs exhibited permanent changes of both forward and reverse gate leakage characteristics for reference HEMTs after the stress. The reverse gate leakage increased more than two orders of magnitude.

PAGE 114

114 Figure 75 shows the measured off state breakdown voltages of reference and protonirradiated InAlN/GaN HEMTs. The breakdown voltage increased from 100 V in the reference device to 160 V in the protonimplantation one, which was irradiated with an energy of 5 MeV and dose of 51015 cm2. It was previously reported that the degradation in DC characteristics after off state stress in GaN based HEMTs was irreversible. Those permanent changes were related to gate contact metal diffusion beneath the gate fingers, along with associated threading dislocation formation, whic h provided extra leakage paths .82 For unirradiated devices, there were many such spots, which were visible as dark fe atures in electroluminescence (EL) spectra and likely the origin of degradation66 and could be related to the metal diffusion into the semiconductor or formation of defects under the gate. However, for those irradiated HEMTs, lots of defects were created during the proton irradiation process. Based on SRIM simulation, the estimated vacancies around 2 DEG cha nnel ranged from 5109 to 21010 cm2 when the conditions of implantation energy of 515 MeV and dose of 51015 cm2 was applied, and an increase in defects that behave as trap sites are expected in the GaN buffer below the channel. These defects were reported to be deep acceptor like traps with high capture cross s ections for both carrier types .174 These traps can capture free electrons and in consequence, the vertical electric field beneath gate metal was i ncreased and extended into the buffer layer. In other words, the depletion mode was modified, increasing the vertical depletion at the expense of lateral depletion. Therefore, the peak electric field in the x direction at drainside gate edge of the irradi ated HEMT was reduced and the reliability of the irradiated HEMT at a similar drain voltage improved as compared to the reference

PAGE 115

115 HEMT. It was reported that higher breakdown voltage was measured in samples wi th more edgetype dislocations .136 In addition, the relationship between breakdown voltage and density of traps formed by threading dislocations was also demonstrated, which concluded that larger trap density l ed to higher breakdown voltage.175 7.1.4 Summary In conclusion, this section demonstrated significant improvement of reliability of InAlN/GaN HEMTs exposed to proton irradiation. The critical voltage for off state electrical step stress was increased from ~+50V to above +100 V. Minimal changes of gate and drain I V characteristics were observed for the HEMTs post proton implantation. The large change in critical voltage was tentatively attributed to the modification of depletion mode under gate, which increases the tendency for vertical depletion instead of lateral depletion. Hence, the peak electric field in the x direction at the drainside gate edge was reduced.

PAGE 116

116 7.2 The Effects of Proton Energy on the Degradation of AlGaN/GaN High Electron Mobility Transis tors 7.2.1 Background AlGaN/GaN high electron mobility transistors (HEMTs) are well suited for high power and high frequency broadband communication systems either on the ground or in space. High radiation resistance is required for applications in satelli tes and space technology because of the presence of large fluxes of highenergy electrons, protons and heavy ions. The initial work on effect of proton irradiation on GaN based heterostructures involved light emitting diodes,176 while subsequent work has focused on AlGaN/GaN HEMTs.165167, 177182 For a proton fluence of 1014 cm2 at 1.8 MeV energy, reductions of saturation drain current (IDSS) and transconductance (gm) in HEMTs from 260 to 100 mA/mm and from 80 to 26 mS/mm, respectively, were reported.177 Similar proton energy studies at different energies were performed by Hu et al.165 and White et al. ,178 which found little degradation and good radiation tolerance of the device channel at fluences up to 1014 cm2. For proton irradiation energy of 5 MeV and doses of 2 1015 cm2, which is equivalent to roughly 1000 years in low earth orbit, the IDSS of AlGaN/GaN HEMTs was decreased by 43%,166 while at 17 MeV at doses of 2 1016 cm2, the reduc tions were 43% and 29% in IDSS and gm, respectively.180 To simulate the environment in space, Sonia and co workers also irradiated devices with 2 MeV protons, carbon, oxygen, iron and krypton ions with fluences ranging from 1 109 cm2 to 1 1013 cm2.181 The energy dependence of protoninduced degradation was studied by Hu et al. ,182 little degradation was observed at 15, 40 and 105 MeV, while 10.6% and 6.1% reductions of drain saturation current and maximum transconductance were obtained at 1.8 MeV energy and fluences of 1012 cm2, due to much larger non

PAGE 117

117 ionizing energy loss .183 184 Roy et al.185 studied the radiation response of GaN/AlGaN HEMTs grown by Molecular Beam Epitaxy (MBE ) to 1.8 MeV proton fluences. HEMTs grown under ammonia rich conditions were more susceptible to protoninduced degradation, compared to devices grown under Ga rich or N rich conditions. Proton irradiation caused positive shifts in pinchoff voltage f or all three kinds of devices. N vacancies were suggested to be responsible for an increase of 1/f noise after irradiation.186 To understand the energy dependence of irradiationinduced degradation in AlGaN/GaN at higher fluence, a proton source with energies of 5, 10 and 15 MeV at a fixed fluence of 5 1015 cm2 was employed in this study The measurement of HEMT gate and drain I V characteristics as well as small and large signal rf measurements were conducted prior to and after the proton irradiation. The dependencies of mobility and carrier concentration on irradiation energy were also investigated. 7.2.2 Experimental The AlGaN/GaN HEMTs were grown on 6H SiC semiinsulating substrates by metal organic chemical vapor deposition, with the following sequence of epitaxial layers: an AlN nucleation layer, a 1.8 m GaN buffer layer, a 1 nm In0.10Ga0.90N backbarrier, a 15 nm GaN channel and a top 22 nm Al0.26Ga0.74N barrier. On wafer Hall measurements showed a sheet carrier concentration, sheet resistance, and mobility of 9.1 1012 cm2, 2/Vs, respectively. An inductively coupled plasma mesa etch of ~1000 was performed to isolate adjacent devices. Ti/Al/Ni/Au Ohmic metallization patterned by lift off was annealed at 850C for 30 sec for source/drain contacts, producing a specific c The HEMTs employed a Ni/Au design with a gate length of 0.2 m and width of 50 m. The source to -

PAGE 118

118 gate and gateto drain distances were 1 m. The devices were passivated with 200 nm of SiNx deposited by a Plasma Therm 790 PECVD system. Figure 7 6 shows a schematic of the AlGaN/GaN HEMT. The dc characteristics of the HEMTs were measured with HP 4156 parameter analyzer. Off state drainvoltage stepstress was also performed on the reference and proton irradiated samples. The HEMTs were stressed for 60 sec onds at each drain voltage step, while grounding the source electrode and applying a constant 6V to the gate electrode. The stress started at 5 V of drain voltage and the voltage step was kept at 1 V. During the step stress, a number of parameters were measured, including gate current, IG, gate to source leakage current, IGS, and gateto drain leakage current, IGD. Between each step stress cycle, the drain I V characteristics, extrinsic transconductance, gate forward current biased from 0 to 1.5 V and gate reverse current biased from 0 to 10 V were also recorded. The RF performance of the HEMTs was characterized with an HP 8722C network analyzer. Load pull measurements were conducted with a Maury microwave system at 10 GHz at room temperature. Stopping and range of ions in matter (SRIM) simulations were used to estimate the penetration depth of the protons into the AlGaN/GaN HEMT structures at various proton energies. 7.2.3 Results and Discussion As shown in Figure 7 7 A and 7 7 B the SRIM data indicate that the majority of the nuclear stopping damage induced by the high energy protons occurs deep in the substrate, 105, 335, and 672 m for 5 10 and 15 MeV, respectively. The two dimension electron gas channel (2DEG) of the HEMT is located 22 nm below the sample surface and the vacancy densities at the 2DEG are several orders low er than the peak of the

PAGE 119

119 damage. The energy loss of protons is primarily due to stopping until near the endof range. As shown in Figure 2a, the 15 MeV protons can penetrate the entire A lGaN/GaN/SiC wafer, which is approximately 500 m thick. For the deeper the proton penetration, the less displacement damage is generated around th e 2DEG, as shown in Figure 77 C Therefore, based on SRIM analysis, 5 MeV protons should degrade the AlGaN/Ga N HEMT more severely as compared with higher energy protons (10 and 15 MeV) if the main degradation mechanism is related to creation of defects by nuclear stopping. Figure 7 8 A shows transfer characteristics from the HEMTs before and after irradiation at 1 0 MeV. The extrinsic transconductance, gm, was reduced by 22 % and there was a positive shift of 0.34 V for the threshold voltage, Vth. These changes were mainly due to the displacement damage induced by the ion bombardment reducing both the carri e r densit y and electron mobility .165 As shown in Figure 7 8 B more severe degradation of the gm and a larger positive Vth shift were observed on HEMT s irradiated with a lower energy of 5 MeV, with the gm decreased around 40% and Vth shifted by almost 1V. The non ionizing energy loss in the actual 2DEG region increased with decreasing proton energy and thus more trap states and scattering centers were created around the AlGaN/GaN interface.183, 184 A similar trend was observed for the effect of proton energy on extrinsic transconductance reduction, with 38 % and 22 % reductions after 5 and 10 MeV irradiation, respectively, and less than 12% for 15 MeV. There was a positive shift of threshold voltage of 0.98 V after 5 MeV proton irradiation, while the HEMT irradiated with a 15 MeV protons exhibited a much smaller shift of 0.05 V.

PAGE 120

120 The reverse and forward gate I V characteristics of the HEMTs before and after proton irradiation at 10 MeV are illustrated in Figure 7 9 The reverse gate leakage current at VG = 10V decreased more than one order of magnitude and similar results were observed for the HEMTs irradiated with 5 and 15 MeV protons. The gate forward characteristics of the proton irradiated HEMTs also improved, and the Schottky barrier height (SBH) increased around 58 % while ideality factor decreased 3040%, as summarized in Table 7 1 Besides the gate forward and reverse characteristics showing significant improvements, there were t wo other features related to the Schottky gate that were also enhanced, namely gate critical voltage during the off state drainvoltage stepstress and off state drain breakdown voltage, as shown in Figure 7 10 A a nd B The critical voltage, Vcri, of the of f state step stress was defined as the onset of a sudden IG increase during the stress.187 During the off state stress, before reaching the Vcri, there w as no degradation observed for both gate and drain I Vs. Once the drain bias voltage reached Vcri, not only did the gate current suddenly increased, but the HEMTs showed permanent degradation involving much higher reverse bias gate leakage current and forw ard gate current, and reductions in Schottky barrier height, drain saturation current, transconductance and drain current onoff ratio .142, 187 The typical Vcri of the reference HEMTs was around 12 to 15 V, by sharp contrast, those of the proton irradiated HEMTs ranged from 45 to 50 V. Thus, the gate electrode of the proton irradiated HEMTs coul d sustain 50V ( 6V applied on the gate during the stress plus 45V of drain voltage) as compared to 18V for the reference samples. As illustrated in Figure 7 10B the off state drain breakdown voltage of the proton irradiated HEMTs als o increased significantly. The gate electrode electrical field did not evenly distribute around the gate, the highest

PAGE 121

121 electrical fields were located at the edges of the gate .82 A possible explanation is that the proton induced defects that formed a virtual gate in the buffer layer, changing the gate electrical field profile of the gate a nd reducing the maximum field. Similar results of improving drain breakdown voltage were also reported for the HEMTs grown with lower quality buffer layer .136, 175 It was suggested that the charged traps in the low quality buffer layer modified the depletion region by increasing the vertical depletion at the expense of decreasing lateral depletion. Tab le 7 2 summarizes the effect of proton irradiation energy on saturation drain current, subthreshold drain leakage current, drain current onoff ratio and sub threshold slope. Subthreshold leakage current, subthreshold slope and on/off drain current rati o are essential to power amplifier performance in power added efficiency, linearity, noise figure and reliability. Although the change in saturation drain current was inversely proportional to the proton energy as a result of trap formation reducing the fr ee carrier density in the HEMT channel, other properties, such as subthreshold drain leakage current, drain current onoff ratio and subthreshold slope, improved significantly as a result of t he lower gate leakage current. Figure 7 11 shows drain and gat e currents as a function of the gate voltage prior to and post 10 MeV proton irradiation. The drain current was slightly reduced, while the subthreshold drain leakage currents were reduced 2 orders of magnitude and the drain current on/off ratio increased mo re than 2 orders of magnitude. The subthreshold slopes also decreased 40% and were closer to the ideal theoretical number of ~60 mV/dec. These subthreshold characteristics were highly dependent on the reverse bias gate leakage current, and significant improvements were observed after the proton irradiation.

PAGE 122

122 Figure 7 12 A shows the drain current voltage (I V) characteristics of AlGaN/GaN HEMTs before and after ir radiation with 10 MeV protons. Although all the irradiated AlGaN/GaN HEMTs exhibited good pinch off characteristics, the amount of saturation drain current reduction was dependent on the irradiation energy. For the 10 MeV irradiated HEMTs, the reduction of saturation drain current at VG = 0V was 24 %. A much larger saturation drain current reducti on, 46 %, was observed for the HEMTs irradiated with5 MeV protons energy and only 11.5% drain current reduction for the HEMTs irradiated with 15 MeV protons, as illustrated in Figure 7 12 B The effect of proton irradiation on the saturation drain current was consistent with the transfer characteristics results. The drain I V characteristics in the low field linear region were used to extract the electron mobility by treating the undoped AlGaN lay er as a gate insulation layer. There were 34 78% reductions in electron mobility observed for the irradiated HEMTs, dep ending on the proton energy. The electron mobility, carrier concentration and sheet carrier concentration, as well as the carrier removal rate of the proton irradiated H EMTs are summarized in Tabl e 73 C V measurements were carried out to estimate the carrier co ncentrations of these samples. The carrier removal rates were defined as the ratio of carrier concentration decrease divided by the fluence of irradiated protons. As illustrated in Figure 7 13 the carrier removal rate was inversely proportional to the irradiated proton energy, RNC = 21.5E (MeV) + 441.7, where RNC is carrier removal rate and E is proton energy. Gate pulse measurements were employed to evaluate traps created during proton irradiation.188 In this technique, the response of the drain current (IDS) to a pulsed gatesource voltage (VGS) was measured. The normalized IDS as a function of VGS in

PAGE 123

123 both pulsed and dc modes for HEMTs before and after 5 MeV proton irradiation at a fluence of 5 1015 cm2 is shown in Figure 7 14A In this case, the VGS was pulsed from 5V to the values shown on x axis at different frequencies of 100, 10 K and 100 KHz with 10% duty cycle, while the drain voltage was kept constant at +5V. The reduction of the drain current in the pulsed mode as compared to the drain current in the dc mode was due to the presence of surface traps in the access region between gate and drain contacts. The comparison of IDS reduction in pulsed mode after irradiation at various energies is illustrated in Figure 7 14B Larger gatelag was produced at higher frequency for HEMTs irradiated at 5MeV, which indicated more shallow traps close to device surface produced at lower irradiation energy. Double pulse measurements were also performed ,189 in which the drain was pulsed from +10 to +5V, while simul taneously pulsing the gate bias from 5V to the values shown on the x axis at different frequencies of 100, 10 K and 100 KHz with 10% duty cycle. As illustrated in Figure 7 15 A the di spersion between dc and pulsed data at gate bias close to threshold voltage was due to the formation of an virtual gate resulting from the injection of hot electrons into the surface between gate and drain elect rodes during off state biases. The normalized double pulsed IDS as a function of irradiation energy is shown in Figure 7 15 B A similar trend was observed in the gatelag result, as a result of more defects created close to the surface in the access region between gate and drain after lower energy ir radiations. Besides DC characterization, small and large signal rf measurements were also performed. The small signal measurements on proton irradiated HEMTs were performed from 50 MHz to 40 GHz. Figure 7 16A shows the microwave performance of

PAGE 124

124 devices exposure to 15 MeV proton irradiation, with the HEMTs biased around the gm peak and VDS = +5V. fT and fMAX can be extracted from the extrapolation of h21 and of Masons maximum unilateral gain U with a slope of 20 dB/dec. fT, fMAX and G as a function of irradiation energy were plotted in Figure 7 16 B Proton irradiation decreased both fT and fMAX, and lower energy proton irradiations produced more degradations in rf characteristics. Load pull measurements were conducted at 10 GHz. Both load and source tuners w ere tuned to optimum states for the device under test (DUT), RF input power was swept from 0 to 15 dBm to record load pull characteristics, as shown in Figure 7 17A The output power (Pout) and power added efficiency (PAE), Gt and Gp referred in Figure 7 1 7 B were measured at peak of PAE. As with the dc and rf results, HEMTs irradiated with lower energy showed more degradation in power performances due to more severe protoninduced displacement damage induced in the HEMT channel at lower irradiation energy. 7.2.4 Summary The effect of proton irradiation energies at 5, 10 and 15 MeV at fixed fluence of 5 1015 cm2 has been studied with dc, rf and power measurements in section 7.2. After irradiation, subthreshold drain leakage current and reverse gate I V d ecreased more than one order of magnitude for all cases due to the increase of resistivity of the HEMT channel. The increase in device degradation with decreasing proton energy is due to the increase in linear energy transfer and corresponding increase in nonionizing energy loss with decreasing proton energy in the active region of the HEMTs.

PAGE 125

125 7.3 The Effects of Proton Dose on the Degradation of AlGaN/GaN High Electron Mobility Transistors 7.3.1 Background The properties of the AlxGa1 xN material system, such as large bandgap, high electron mobility and breakdown field, make it a promising candidate for applications in high power and high frequency communication systems.190192 For space based applications, electronic components of satellites may suffer radiation damage from high fluxes of energetic particles in the van Allen belts, such as protons, electrons and heavy ions caused by solar flares and primary cosmic rays.193 As compared to GaAs, GaN demonstrates several orders of magnitude higher radiation tolerance. Recently, attention has been focused on the effects of proton irradiation of GaN based high electron mobility transistors (HEMTs) at energies in the range 515 MeV and relatively high doses of 5 1015 cm2.194, 195 These results have shown larger degradation of dc (saturation drain current, IDSS and transconductance, gm) and rf (unit gain cutoff frequency, fT, maximum oscillation frequency, fmax and power added efficiency, PAE) characteristics for HEMTs irradiated at the lower range of these proton energies because of the higher nuclear stopping energy loss of these protons at the shallow depths around the 2dime ntional electron gas channel (2DEG). Hu et al. studied the effects of proton energy with a wide range of 1.8, 15, 40 and 105 MeV at a fluence up to 1013 cm2.196 In this case, the devices exhibited little degradation when irradiated with 15, 40, and 105 MeV protons, while the greatest degradation was measured at the lowest proton energy, due to the larger nonionizing energy loss of the 1.8 MeV protons .196 The effects of proton dose on dc characteristi cs of InAlN/GaN HEMTs were

PAGE 126

126 reported by Lo et al. ,197 which revealed that more degradation was induced at higher irradiation dose, with reductions of gm of 1% and 15%, an d increase of channel resistance of 6% and 28% for HEMTs exposed to 21011 and 21015 cm2 protons, respectively. In addition, it was also reported that there was little degradation at doses below 1014 cm2 in IIInitride HEMTs .165 The proton radiation introduces point defects in the GaN, which can decrease sheet carrier mobility due to increased carrier scattering and decreased sheet carrier density by carrier removal.165 Experimental results demonstrate several types of defects after proton radiation, including N vacancies (VN d onor) near Ec (0.04 0.06) eV,198 Ga vacancies (VGa acceptor) near Ev + 1 eV ,199 nitrogen interstitials (Ni acceptors) near Ec 1 eV and Ga interstitials (Gai donors) near Ec 0.8 eV.200, 201 Recently the accompanying improvement of the reliability of protonirradiated AlGaN/GaN HEMTs at energies of 5, 10 and 15 MeV was report ed.195 In this chapter, an investigation of the effects of radiation dose on dc ch aracteristics of irradiated HEMTs was reported. The AlGaN/GaN HEMTs were irradiated with 5 MeV protons at doses ranging from 1109 to 21014 cm2. The dependencies of mobility, sheet carrier concentration and carrier removal rate on irradiation energy were also investigated. 7.3.2 Experimental The AlGaN/GaN HEMT device structures were grown on semi insulating 6H SiC substrates and consisted of a thin AlN nucleation layer, 2.25 m of Fedoped GaN buffer, 15 nm of Al0.28Ga0.72N, and a 3 nm undoped GaN cap. Onwafer Hall measurements showed sheet carrier concentrations of 1.061013 cm2, mobility of 1907 cm2 /V isolation, Ti/Al/Ni/Au Ohmic contacts alloyed at 850o

PAGE 127

127 and dual finger Ni/Au gates patterned by lift off. The gate length was 1m, and gate width was 2150 m. Both source to gate gap and gateto drain distances were 2 m. The devices exhibited typical maximum drain currents of 1.1 A/mm, extrinsic transconductance of 250 mS/mm at VDS of 10 V, threshold voltage of 3.6 V. The devices were passivated with 200 nm of SiNx deposited by a Plasma Therm 790 PECVD system. Proton irradiations were performed at the Korean Institute of Radiological & Medical Sciences (KIRAMS) using a MC 50 (Scanditronix) cyclotron. The proton energy at the exit of the cyclotron was 30 MeV. The proton energy at the sample was 5 MeV after passing through two aluminum degraders. The thickness of each aluminum degrader was 2.7mm. The beam currents were measured using Faraday cup to calculate flux density. In this study, the proton dose was varied from 1109 to 21014 cm2. Stopping and Range of Ions in Matter (SRIM) simulator was used to estimate the penetration depth of the protons into the AlGaN/GaN HEMT structure. The device DC characteristics and device reliability test were performed with a HP 4156 parameter analyzer. 7.3.3 Results and Discussion Figure 7 1 8 illustrates the sheet resistance (RS), contact resistivity (RC) and transfer resistance (RT) of the AlGaN/GaN HEMTs extracted from transmission line measurements (TLM) prior t o and post proton irradiation as a function of proton dose. For the dose of 109 cm2, there was no change for all those parameters. The threshold of RC and RT degradation was at a proton dose of 5109 cm2, and the RC and RT increased linearly proportional to the proton dose until the proton dose reached at 21013 cm2, exhibiting 3 and 5.5% increases for RT and RC, respectively. However, the

PAGE 128

128 threshold of the proton dose for RS degradation was much higher at 21013 cm2 as compared to those for RT and RC. There was no degradation for RS detected for the lower dose protons irradiation, in which the density of irradiationinduced defects was still negligible as compared to the native defect density. For the condition of the highest proton dose of 21014 cm2 used in this study, RS, RC and RT increased 7.9, 6.7 and 7.5%, respectively. The increase of RT and RC under the Ohmic contacts could be due to more defects created by proton irradiation in these disordered regions. It is well known that significant increases of edgeand mixedtype threading dislocations (TDs) are induced by metal contact inclusions after high temperature (>850C) Ohmic contact annealing.202 Similar trend was reported by Karmarkar et al. .203 Ohmic contact metal regions were more prone to proton irradiation damage, consistent with SRIM simulations shown in Figure 7 19 Figure 7 19A shows the SRIM simulation of ion energy loss as a function of proton penetration depth into the HEMT structure grown on the SiC substrate. The majority of the energy loss was through nuclear stopping deep in the SiC substrate around 145 m below the HEMT structure. The simulated penetration depth of 5 MeV protons was around 150 m, and Ohmic metal contact was too thin to affect the penetration. Due to light mass of protons, the nuclear energy loss was minimal in the HEMT surface region, where energy loss was dominated by the electronic stopping mechanism, as illustrated in Figure 7 19B The 2dimensional electron gas (2DEG) region is located around 2530 nm below the surface, thus, there was no damage detected for RS until the proton dose reached 21013 cm2. However, as illustrated in Figure 7 19 C more energy loss w as in the Ohmic metal stack region due to nuclear stopping of protons with heavier mass Au atoms in the Aubased Ohmic

PAGE 129

129 metallization. Those scattered protons from the collisions with Au atoms could damage nearby lattices in the 2DEG region. Thus RC and RT showed a tendency to be more affected by high energy irradiation than RS. To simplify the simulation, the scheme of the Ohmic metallization prior to the high t emperature annealing was used. After annealing, the top Au layer would diffuse to the metal/AlGaN interface, which could induced more proton scattering and create even m ore defects in the 2DEG channel .204 The changes of reverse gate leakage current, threshold voltage, and extrinsic transconductance, gm, were within the measurement errors f or the HEMTs irradiated with a proton dose less than 21013 cm2, as shown in Table 74 Figure 7 20 illustrates drain and gate currents as well as the typical transfer characteristics of HEMTs as a function of gate voltage prior to and after 21014 cm2 p roton irradiation, respectively. These measurements were conducted at a fixed drain voltage of +5V. As shown in Figure 7 20 A the subthreshold drain leakage current was dominated by the reverse gate leakage current when the channel was pinchedoff. There was minimal change of reverse gate leakage current, while there was a reduction of extrinsic transconductance (gm) of 10% and a positive shift of threshold voltage (Vth) of 95 mV after irradiation, as shown in Figure 7 20B These degradations could be att ributed to the displacement damage, resulting in the reduction of carrier concentration and mobility9. Figure 7 21 illustrates the drain I V characteristics of AlGaN/GaN HEMTs before and after proton irradiations with different doses, measured with VG sta rtin g from 0 V with a step of 1V. The IDSS of HEMTs irradiated with a dose less than 21013 cm2 showed minimal change, however, a degradation of 13% was observed when a higher dose of 21014 cm2 was employed. Table 75 summarizes the IDSS, the reduction of

PAGE 130

130 2DEG mobility, sheet carrier concentration and carrier removal rate as a function of proton irradiation dose. Hall measurements were used to determine the 2DEG mobility and sheet carrier concentrations were estimated from these electron mobilities and the sheet resistances measured with TLM. The carrier removal rates were defined as the ratio of carrier concentration decrease divided by the fluence of irradiated protons. There were no apparent changes of mobility and sheet carrier concentration for thos e H EMTs irradiated with low doses. The HEMTs irradiated with a dose less than 21013 cm2 showed minimal changes of mobility and sheet carrier concentration, however, decreases of 10% and 41% in sheet carrier concentrati on and mobility, respectively, were observed for the HEMTs exposed to 21014 cm2 proton dose. The carrier removal rate was determined to be 810 cm1. In contrast to the trends in IDSS, HEMTs irradiated with higher doses of protons exhibited higher drain breakdown voltages, VBR, as illustrat ed in Table 7 6 Below a threshold dose of 21013 cm2, the VBR was fairly constant around 30 1 V. However, the VBR of HEMTs irradiated with higher doses increased by 20% and 37% for the HEMTs irradiated at 21013 cm2 and 21014 cm2 protons, respective ly. The VBR was highly dependent on the electrical field distribution around gate edges. The electrode electrical field does not evenly distribute around the gate, with the highest electrical fields located at the edges of the gate and field plate used to reduce the peak electrical field on the edges of the gate electrode.82 It was proposed previously that proton induced defects form a virtual gate in the buffer layer, changing the gate electrical field profile of the gate and reducing the maximum field.195 Similar results of improving drain breakdown voltage were also reported for the HEMTs grown with lower quality buffer

PAGE 131

131 layers.136, 205, 206 It was suggested in that work that the charged traps in the low quality buffer layer modified the gate depletion region by increasing the vertical depletion at the expense of decreasing lateral depletion. Rather than creating the defects in the GaN buffer layer by epitaxial growth as in past reports, in our work the high energy proton irradiation was used to create defects in the GaN layer and enhance t he drain breakdown voltage. To illuminate the trap characteristics, gate pulse measurement was conducted. In this study, the drain current (IDS) response to a pulsed gatesource voltage (VGS) was measured. The normalized IDS as a function of VGS in both pulsed and dc mode for the HEMT prior to proton implantation is shown in Figure 7 22A During the pulse measurement, the VGS was pulsed from 5V to the values shown on x axis at frequencies of 100Hz, 10KHz and 100KHz with a duty cycle of 10%, while drain was kept constant at + 5V. The reduction of the drain current in the pulsed mode as compared to the drain current in the dc mode was due to the presence of surface traps in the access region between gate and drain contact. Usually, the traps have some specific time constants and could not respond above certain frequency. Thus, the pulsed drain current would be lower than the dc drain current. As shown in Figure 7 22B there was apparent reduction of IDS for the HEMTs irradiated at proton doses higher than 2101 2 cm2, which meant more traps were introduced for the implanted HEMTs. Drain pulse measurements were also performed on the reference and proton irradiated HEMTs by pulsing the drain from 0 to different voltages up to 15 V. Figure 7 23A and B illustrate no rmalized dc and pulsed drain current at three different frequencies for the reference and HEMT irradiated with a dose of 21014 cm2 protons

PAGE 132

132 respectively. Both the drain current measured at dc mode for both HEMTs exhibited 20% reduction due to heating ef fect, but not for the pulsed measurements. Figure 7 23 C shows the drain current reduction of HEMTs as a function of proton implantation dose at different frequencies. Similar trend was observed as in gate pulse measurements that more traps generated for the HEMTs irradiated with higher doses of the protons. Higher reduction of the drain current was detected with higher pulse frequency due to slow response of the traps as compared to the pulsed drain voltage. Off state drainvoltage stepstress was also cond ucted on AlGaN/GaN HEMTs prior to and after proton irradiation to evaluate device reliability .21 The HEMTs were constantly biased for 60 s at each drainvoltage step, while grounding the source electrode and fixing the gate voltage at 8V. The stress started at 5 V of drain voltage and drain voltage step was 1 V during electrical stress. Figure 7 24 shows gate current, IG, as a function of stressed drain voltage. During the stepstress, besides monitoring IG, gateto source leakage current, IGS, and gateto drain leakage current, IGD, were also measured. Due to the low drainto source current during off state, self heating effect was negligible and had no effect on device performance. The critical voltage, Vcri, of the off state step stress was defined as the onset of a sudden IG increase during the stress. As previous work reported, permanent damage in the devices may be created upon exceeding Vcri, including the formation of cracks on both source and drain sides of gate edges, the diffusion of gate metal and native oxide layer at the interface between metal and AlGaN barrier layer, as well as associated threading dislocations, all of which could provide possible leakage paths.40, 82 These mechanisms also cause the irreversible degradation

PAGE 133

133 of dc characteristics of GaN based HEMTs .21, 204 Ther e was no difference in Vcri, 22 1 V, detected for the pristine and protonirradiated HEMTs with a dose less than 21013 cm2, as shown in Table 76 However, larger Vcri values of 28 and 32 V were observed for HEMTs irradiated with higher doses of 21013 and 21014 cm2, respectively. The improvement of Vcri could be ascribed to the previously described mechanism for the improvement of drain breakdown voltage.195 A virtual gate was formed in the buffer layer for the proton irradiated HEMTs, which reduced the maximum electric field by extending the depletion region into the buffer layer. 7.3.4 Summary In this section, t he effects of proton doses on dc characteristics of irradiated AlGaN/GaN HEMTs by dc measurement and off state electrical stress was reported. There were less degradation in saturation drain current (IDSS), transconductanc e (gm), mobility and sheet carrier concentration at doses below 2 1013 cm2. As irradiation dose increased, the increase of VBR were 20% and 37%, Vcri were 27% and 45% at doses of 2 1013 and 2 1014 cm2, respectively. The improvements of VBR and Vcri could attribute to the modification of depletion region due to the introduction of a higher density of defects after irradiation at a higher dose.

PAGE 134

134 Figure 71. Off state gate currents as a function of drain voltage for the unirradiated and protonirradiated HEMTs. The protons had energy of 5 and 10 MeV and dose of 51015 cm2. 0 20 40 60 80 100 10-510-410-310-210-1100101 IG (A) Reference 5 MeV 10 MeV IG (mA/mm)VDS (V) VG = -6V10-810-710-610-510-410-3

PAGE 135

135 Figure 72. Drain I Vs of un irradiated HEMTs prior to and after off state electrical stepstress. The devices were stressed with VG = 6 V for 60 s at each drain voltage step until sudden increase of IG was observed. 0 2 4 6 8 10 0 200 400 600 800 1000 before stress after stressIDS (mA) IDS (mA/mm)VDS (V)VG = 0V, in -1 stepUn-irrad.0 50 100 150 200

PAGE 136

136 Figure 73. Drain I Vs of HEMTs irradiated with 5 MeV and 51015 cm2 doses of protons prior to and after off state electrical stepstress. The devices were stressed with VG = 6 V for 60 s at each drain voltage step until drain voltage reached +100V. 0 1 2 3 4 5 0 100 200 300 400 500 before stress after stressI DS (mA) I DS (mA/mm) VDS (V) VG = 0V, in -1 stepIrrad. 0 20 40 60 80 100

PAGE 137

137 Figure 74. Gate characteristics of unirradiated and protonirradiated (with 5 MeV and 51015 cm2 dose) HEMTs prior to and after off state electrical step stress. Un irradiated devices were stressed with VG = 6 V for 60 s at each drain voltage step until sudden increase of IG was observed. The same condition used for irradiated HEMT except the drain voltage reached +100 V. -10 -8 -6 -4 -2 0 2 10-810-610-410-2100 Un-irrad. Irrad. before stress after stressIGS (A) IGS (mA/mm)VGS (V)10-1010-810-610-4

PAGE 138

138 Figure 75. Off state breakdown measurement result of unirradiated and protonirradiated HEMTs. The incident protons had energy of 5 MeV and dose of 51015 cm2. 0 50 100 150 200 0 20 40 60 80 100 Current ( A)VDS (V) Virgin HEMT Irrad. HEMTVG = -6V

PAGE 139

139 Figure 76 A schematic of the AlGaN/GaN high electron mobility transistors.

PAGE 140

140 Figure 77 SRIM simulation results at the AlGaN/GaN interface of the AlGaN/GaN HEMT structure. A ) Energy loss and B ) Vacancy density as a function of target depth C ) Vacancy densitie. 0 200 400 600 800 1000 1200 0 2 4 6 8 10 Energy loss (eV/Ang.)Target depth ( m) 5 MeV 10 MeV 15 MeV(A) 0 200 400 600 800 1000 1200 0 2x10-54x10-56x10-58x10-51x10-4 (B)Target depth ( m)Vacancies (No./Ang.-Ion) 5 MeV 10 MeV 15 MeV 10-710-610-5 (C) Vacancies (No./Ang.-Ion)15 MeV 10 MeV 5 MeV

PAGE 141

141 Figure 78 D C ch a r a c t e r i st i cs of HEMTs pre and post proton irradiation with energy of 10 MeV at a fluence of 5 1015 cm2. A ) Transfer characte ristics B ) Threshold voltages shift and transconductance reduction as a function of irradiation energies. -6 -5 -4 -3 -2 -1 0 0 200 400 600 800 1000 1200 E = 10MeV G m (mS/mm) IDS (mA/mm)VG (V)Fresh Irrad. IDS IG VDS = +5V 0 200 400 600 (a) (A) 0.0 0.2 0.4 0.6 0.8 1.0 (b) Transconductance reduce (%)15 MeV 10 MeVVT shift (V)5 MeV 0 10 20 30 40 (B)

PAGE 142

142 Figure 79 Gate IV of HEMTs pre an d post proton irradiation with an energy of 10 MeV at a fluence of 5 1015 cm2. -10 -8 -6 -4 -2 0 2 10-710-510-310-1101 IG (A)E = 10MeVIG (mA/mm)VG (V) Fresh Irradiation10-1010-810-610-4

PAGE 143

143 Figure 710 Off state drainvoltage a n d o ff state drain breakdown voltage of HEMTs p re a nd post proton irradiation. A ) Off state drainvoltage stepstress gate current as a function of drain voltage. B ) Off state drain breakdown voltage of HEMTs pre a nd post proton irradiation. 0 10 20 30 40 50 60 70 10-410-310-210-1100101 (A) Un-irradi. 5 MeV IG (mA/mm)VDS (V) VG = -6V 0 5 10 15 20 25 30 0.0 0.2 0.4 0.6 0.8 1.0 1.2 (B) Un-irradi. 5 MeVCurrent ( A)VDS (V) VG = -8V

PAGE 144

144 Figure 711 IDS and IG as a function of VG of HEMTs preand post proton irradiation with an energy of 10 MeV at a fluence of 5 1015 cm2. -6 -5 -4 -3 -2 -1 0 10-510-310-1101103105 IDS (A) IDS (mA/mm)VG (V)Fresh Irrad. IDS IGVDS = +5V E = 10MeV 10-810-610-410-2100

PAGE 145

145 Figure 712 Drain I Vs of HEMTs pre and post proton irradiation with an energy of 10 MeV A ) Drain I Vs B ) Saturation current at VDS = +5V as a function of irradiation energies. 0 1 2 3 4 5 0 200 400 600 800 1000 1200 (a)IDS (mA)E = 10MeVIDS (mA/mm)VDS (V) Fresh Irradiation0 10 20 30 40 50 60 (A) 0 20 40 60 (b)15 MeV 10 MeVDrain Current Reduction (%)5 MeV (B)

PAGE 146

146 Figure 713 Carrier removal rate as a function of proton irradiation energies at a fixed fluence of 5 1015 cm2. 5 10 15 100 200 300 400 RNC = -21.5E(MeV) +441.7Carrier Removal Rate, RNC (cm-1)Proton Energy (MeV)

PAGE 147

147 Figure 714 T h e r e s u l t o f G a t e p u l s e m e a s u r e m e n t A ) Gate pulse measurements on HEMTs pre and post proton irradiation with energy of 10 MeV at a fluence of 5 1015 cm2. VG is switched from 5V to the values shown on the x axis at a 10% duty cycle. B DS as a function of irradiation energies. -5 -4 -3 -2 -1 0 1 0 20 40 60 80 100 (A)E=10MeV DC result 100Hz 10KHz 100KHz VDS = +5VNormalized IDS (%)VG (V) 0 5 10 15 20 (B) 100Hz 10KHz 100KHz15 MeV 10 MeV IDS (%)5 MeV

PAGE 148

148 Figure 715 T h e r e s u l t o f d o u b l e p u l se m e a s u r e m e n t A ) Double pulse measurements on HEMTs preand post proton irradiation with energy of 10 MeV at a fluence of 5 1015 cm2. Drain was pulsed from +10 to +5V, simultaneously pulsing t he gate from 5V to the values shown on the x axis at a 10% duty cycle. B DS as a function of irradiation energies. -5 -4 -3 -2 -1 0 1 0 20 40 60 80 100 (A)E=10MeV DC result 100Hz 10KHz 100KHz VDS = +5VNormalized IDS (%)VG (V) 0 20 40 60 80 (B) 100Hz 10KHz 100KHz15 MeV 10 MeV IDS (%)5 MeV

PAGE 149

149 Figure 716 T h e r e s u t l s o f sm a l l si g n a l m e a s u r e m e n t A ) Small signal measurements on HEMTs after proton irradiation with an energy of 10 MeV at a fluence of 5 1015 cm2. B ) fT and fmax as a function of irradiation energies. 0.1 1 10 100 0 20 40 60 80 100 (A)Gain (dB)Frequency (GHz) U H21E=10MeV 20 40 60 80 100 120 (B) fmax fT15 MeV 10 MeVfT, fmax(GHz)5 MeV

PAGE 150

150 Figure 717 T h e r e s u t l s o f l a r g e si g n a l m e a s u r e m e n t A ) Power measurements at 10 GHz on HEMTs after proton irradiation with an energy of 10 MeV at a fluence of 5 1015 cm2. B ) Load pull characteristics as a function of irradiation energies. 0 2 4 6 8 10 12 14 0 10 20 30 40 50 0 10 20 30 40 50 E = 10 MeV Pout Gt Gp PAEPout (dBm), Gt, Gp (dB)Pin (dBm)(A)Efficiency (%) 0 10 20 30 40 50 Efficiency (%) Pout Gt Gp PAE15 MeV 10 MeV 5 MeVPout (dBm), Gt, Gp (dB)0 10 20 30 40 50 (B)

PAGE 151

151 Figure 7 1 8 Percent increases of sheet resistance ( RS), contact resistivity ( RC), and transfer resistance ( RT) after 5MeV proton irradiation with different doses. 1081091010101110121013101410150 2 4 6 8 10 5 MeV Rs Rt RcResistance increase (%)Dose (cm-2)

PAGE 152

152 Figure 719 T h e r e s u l t o f S R I M si m u l a t i o n A ) SRIM simulation of proton ion energy loss and proton penet ration depth into the HEMT structur e grown on the SiC substrate. B ) SRIM simulation of proton ion energy loss and proton penetration depth near the surface of t he AlGaN/GaN HEMT structure. C ) SRIM simulation of proton ion energy loss and proton penetratio n depth in the Ohmic metal contact region of an AlGaN/GaN HEMT. 0 50 100 150 200 250 300 0 10 20 (A)5MeVEnergy Loss (eV/Ang)Target Depth ( m) Without Ohmic metal With Ohmic metal -2000 -1000 0 1000 2000 3000 0 2 4 6 8 (B) Target Depth (Ang.)Without Ohmic metal Energy Loss (eV/Ang.)2-DEGGaNAlGaN 0 2000 4000 6000 0 2 4 6 8 Al Ti GaN AuTarget Depth (Ang.)AlGaNWith Ohmic metal Energy Loss (eV/Ang.)2-DEGNi(C)

PAGE 153

153 Figure 720 D C ch a r a ct e r i st i cs of HEMTs before and after 5 MeV proton irradiation. A ) Drain and gate currents as a function of gate voltage. B ) Typical transfer characteristics. -5 -4 -3 -210-610-410-2100102104 (A) I DS I G (A) I DS I G (mA/mm) VG (V) Fresh Irradi. IDS IGVDS = +5V10-810-610-410-2100 Dose = 2 -5 -4 -3 -2 -1 0 0 200 400 600 800 Dose = 2 G m (mS/mm) I DS (mA/mm) VG (V)Fresh Irradi. Ids GmVDS = +5V (B) 0 200 400 600 800

PAGE 154

154 Figure 721 Drain characteristics of HEMTs prior to and post 5MeV proton irradiation with various doses. 0 5 20 30 40 50 0 200 400 600 800 IDS (mA) VG = 0V, in -1V step Reference 2 1013 cm-2 2 1014 cm-2IDS (mA/mm)VDS (V)0 50 100 150 200

PAGE 155

155 Figure 722 T h e r e s u l t s o f Gate pulse measurements A ) Gate pulse measurements performed on reference HEMT by switching VG from 5V to the values shown on the x axis at different frequencies and a duty cycle of 10%. B ) Drain current reduction as a function of irradiation doses during the gate pulse measurements. -4 -3 -2 -1 0 1 0 20 40 60 80 100 (A) IDS (DC result) 100Hz 10kHz 100kHz VDS = +5VNormalized IDS (%)VG (V)Reference 0 1091010101110121013101440 60 80 100 Proton Dose (cm-2)Reduction of IDS (%) 100Hz 10kHz 100kHz(B)

PAGE 156

156 Figure 723 T h e r e s u l t s o f D r a i n pulse measurements A ) Drain pulse measurements performed on reference HEMT and B ) the irradiated device with a dose of 2 1014 cm2 by s weeping VDS from 0 V to 15V at different frequencies and a duty cycle of 10%. C ) Drain current reduction as a function of irradiation doses during the drain pulse measurements with VDS= 15V 0 5 10 15 0 20 40 60 80 100 VDS (V)Normalized IDS (%) (B) VG=-2 V Dose: 2 10 cm DC 100 Hz 10 kHz 100 kHz 0 5 10 15 0 20 40 60 80 100 (A) VG=-2 V Fresh DC 100 Hz 10 kHz 100 kHzVDS (V)Normalized IDS (%) 0 10 20 30 2 10152 10142 1012 VDS=+15V 100 Hz 10 kHz 100 kHzCurrent Reduction (%)Dose (cm-2)Ref. (C)

PAGE 157

157 Figure 724 Offstate drain stepstress of HEMTs prior to and post 5MeV proton irradiation with various doses. 0 10 20 30 40 50 10-310-210-1100 Ref. 2 1012 cm-2 2 1013 cm-2 2 1014 cm-2VG = -8V IG (A) IG (mA/mm)VDS (V)5 MeV10-610-510-4

PAGE 158

158 Table 71. Summary of IG at VG = 10V, Schottky barrier height before and after 5, 10 and 15 MeV proton irradiation with a dose of 5 1015 cm2. Samples Condition I G at V G = 10V (mA/mm) Schottky barrier height (mV) 5 MeV Pre irrad. 1.3 10 2 730 Post irrad. 4.5 10 4 788 10 MeV Pre irrad. 1.6 10 2 709 Post irrad. 3.3 10 4 740 15 MeV Pre irrad. 1.7 10 2 664 Post irrad. 1.7 10 3 694 Table 72. Summary of the dependence of ON/OFF ratio, saturation drain current, subthreshold drain leakage current and subthreshold slope on proton irradiations. Samples Condition ON/OFF ratio Saturation drain current (mA /mm) Sub threshold drain leakage current (mA/mm) Sub threshold slope (mV/dec) 5 MeV Pre irrad. 6.3 10 4 999 1.6 10 2 179 Post irrad. 1.3 10 6 536 4.2 10 4 159 10 MeV Pre irrad. 4.9 10 4 986 2.0 10 2 202 Post irrad. 1.3 10 6 754 5.6 10 4 124 15 MeV Pre irrad. 4.2 10 4 980 2.3 10 2 190 Post irrad. 9.9 10 5 885 8.9 10 4 172

PAGE 159

159 Table 73. Summary of the dependence of normalized mobility, carrier concentration, sheet carrier concentration and carrier removal rate on the proton i rradiations. Table 74. Summary of threshold voltage sh ift, the reduction of extrinsic transconductance, sheet carrier concentration and mobility, as well as carrier removal rate of HEMTs prior to and post 5MeV proton irradiation with various doses. Irradiation Dose (cm 2 ) th (mV) m (%) Reverse Gate Leakage at VG = 5V and V DS = 5V (A/mm) 510 9 0 0 3.6 510 10 0 0 3.8 210 12 0 0 3.5 210 13 10 5 5.6 210 14 95 10 8.1 Pre irrad. 5 MeV 10 MeV 15 MeV Normalized mobility 1 0.23 0.38 0.66 Carrier concentration (cm3) 5.8 1018 4.3 1018 4.9 1018 5.1 1018 Sheet carrier concentration (cm2) 7.0 1012 5.2 1012 5.9 1012 6.1 1012 Carrier removal r ate (cm1) 336 224 121

PAGE 160

160 Table 75. Summary of IDSS, sheet carrier concentration and mobility, as well as carrier removal rate of HEMTs as a function of proton irradiation doses. Irradiation Dose (cm2) IDSS (mA/mm) Reduction of Sheet Carrier Concentration (%) Reduction of Mobility (%) Carrier Removal Rate (cm1) 510 9 726 0 0 510 10 725 0 0 210 12 725 0 3 210 13 716 1 7 850 210 14 630 10 41 810 Table 76. Summary of the dependence of drain breakdown voltage (VBR) and critical voltage (Vcri) during the off state drainvoltage stepstress as a function of irradiation dose. Irradiation Dose (cm 2 ) Drain Breakdown Voltage VBR (V) Critical Voltage, V cri (V) Pristine 30 22 10 9 31 23 510 9 30 22 510 10 29 22 210 12 31 23 210 13 36 28 210 14 41 32

PAGE 161

161 CHAPTER 8 LASER ABLATION OF S ILICON C ARBIDE 8 .1 Background Silicon Carbide is an at tractive wide energy bandgap semiconductor being used in microelectronics on account of its excellent mechanical strength, thermal conductivity, and breakdown field as microwave power electronics, sensors, actuators, resonators and as a substrate for growi ng GaN based light emitting diodes as well as AlGaN/GaN based highelectron mobility transistors (HEMTs).207211 Vertical throughwafer electrical interconnects (vias) between the metal contacts on the front side of the wafer and the common ground in the back of the wafer are highly desirable to reduce the source inductance, improve thermal conductivity and device reliability.212228 The low inductance vias result in the enhancement of the transistors performance in silicon carbon based electronics, such as SiC metal semiconductor field effect transistors (MESFET s), SiC power metal oxide semiconductor field effect transistors (MOSFETs), and AlGaN/GaN HEMTs grown on SiC substrates. These devices have uses in high power, high temperature commercial applications in telecommunications, hybrid electric vehicles, power flow control and remote sensing.212215 In addition, military applications for RF transmitters and receivers include all weather radar, surveillance, reconnaissance, electronic attack and communications systems. GaN and SiC can potentially operate from VHF through X band frequencies while providing higher breakdown voltage, better thermal conductivity and wider transmission bandwidths than conventional devices It is well known that Silicon Carbide is one of the most difficult machining materials, due to its chemically inert properties involving the high energy Si C bond. Currently, inductively couple plasma etching is widely used to etch vias in the SiC

PAGE 162

162 subst rat e selectivity of the etching mask and timeconsuming preparation required for plasma processing even when the thickness of the silicon carbide substrates is reduced micro masking during ICP etching, which will hinder the subsequent metallization in the via holes.225231 It has reported that the mechanism of the formation of pil lars was redeposition of the nonvolatile etching products NiSiF, generated by a chemical combination of Ni from the commonly used metal mask and SiFx species during etching.229 Although the micromasking effect can be minimized by introducing an Ar pretreatment as well as mixed CF4 w ith SF6/He gases,230 the low etching rate is still an obstacle for efficiency and large scale production of via holes on SiC substrates. So far, the highest etching rate reported is approximate 2 m/min.228 The typical plasma etch rates for 4H and 6H SiC substrates are 0.2ng (1.4 6 hours). There is an interest in developing laser drilling processes for creating throughwafer vias in SiC substrates for devices as an attractiv e alternative to traditional microelectronic fabrication processing, such as wet chemical and plasmabased dry etching.229231 Laser drilling can create via holes in substrates without having a metal mask fabricated on the wafer and in particular for standard thickness substrat es would give added flexibility for creating custom patterns in the substrates through computer control of the laser drilling location and would also eliminate the need for wafer thinning prior to via formation.231 Moreover, there are no pillars formed during laser ablation process. Via holes with diameter 60230 231 In order to increase circuit density, there is a need to form via holes with smaller diameter. It is also possible

PAGE 163

163 to drill the via hole directly under the source cont act metallization of the FETs with the smaller dimension via hole technology, and this would effectively decrease the junction temperature and also the reduce the inductance of the via holes. In this chapter the fabrications of via holes with diameter ran ging from 10 to 70 m and also rectangular via holes in SiC substrate were demonstrated The dependence of drilling rate and the shape of via holes on the diameter of via holes were studied. 8 .2 Experimental 2 in. diameter undoped Silicon Carbide substrat es were used in this work and the samples were drilled with a JPSATM IX convex/convex/convex tripler objective lens with a meniscus corrector was used to correct the spherical aberration and the focal length of t he tripler was 10 cm. A metal mask was installed in the light path and the openings on the mask were imaged onto the target surface with a demagnification of 23. If the same size via holes are drilled, there is no need to change this metal mask. Thus, the steps of standard photolithography and metal deposition of thick Ni layer used to form the dry etch mask can be skipped. However, the laser drill system is equipped with a computer controlled motor driven mask holder to produce via holes of different sizes on the same wafer. 46 different via hole patterns can be fitted on the same metal mask and the patterns can be changed automatically. A schematic diagram of experimental set up is shown in Figure 8 1. In the beam delivery viewing image of the drilled su rface, the CCTV could be automatically adjusted parfocally and coaxially to the focused laser beam spot. The CCTV consisted of a multi element objective lens with coarse/fine focus barrel, a kinematic mirror mount, a CCD high resolution camera with lens and corrective optics, and a 15" monitor with electronic

PAGE 164

164 crosshair. The sample stage was designed to accommodate 6 inch wafers. Two HeNe lasers were used to guide the highaccuracy air bearing x y linear motor sample stage, and the minimum stage movement was 0.1m/step. The stage had an accuracy of +/ 3 8 inches per second; the stage position could be programmed with a resolution of +/ repeatability of +/ The laser pulse duration was fixed at 25 ns and the repetition rate was set at 100 Hz in this study. The energy density of the focused processing beam was in the range 3 12 J/cm2. Microposit 1045 photoresist was coated on the SiC samples prior to the laser drilling as a blank mask to protect the surface from the debris around the drilling zone. The resist was removed after the drilling by dipping the sample in acetone. holes with the 100 100 100 were drilled using circular and rectangular metal masks installed in the light path as shown in Figure 8 1. The shape and depth of the drilled via holes and trenches were studied by dicing across the openings, and then examined with a scanning electron microscopy (SEM). 8 .3 Results and Discussion Figure 8 2 illustrates the drilling rate of 4H SiC as a function of A) opening size and B) diameter via hole, which were around three orders of magnit ude faster than the rate of etching SiC with conventional ICP. A drill rate of 10.2 m/sec corresponds to an ablation rate of 0.1 m/pulse at 100 Hz. When the diameter of the via holes was decreased, the drilling rate increased and reached a maximum at around 18 m/sec for the 10 m via opening employed. The system utilized was a research type equipment

PAGE 165

165 and the maximum focused image of the UV laser light on the drill surface was limited to 320 an unreasonable long drill time, 456 days, to process a 3 wafer. However, there are several higher powered lasers commercially available, such as Coherents LPX Pro series, which can deliver similar fluences with much larger beam spot size, around 1.2 cm 0.5 cm for 450 mJ. With production type systems, only seventy six of 24 second exposures, or 30 minutes, are needed to process a 3 wafer. Thus, multiple via holes can be fitted within the larger field dimension and can be drilled at the same time to enh ance the throughput. In the previous w ork, the drilling rate of via holes were strongly dependent on the opening size of the via hole in glass substrates was observed.231 This drilling rate increase for the smaller via hole was attributed to the reflected laser light from the side wall of the via hole dominating the drilling process. The reflected laser light focused on the bottom of the via hole and enhanced the drilling rate. It was also reported that the diminished screening of the plume, when the size of drilled hole was decrease, also increased the drilling rate.232, 233 The via hole drilling rate increased linearly with the fluence, regardless with the with 50 m opening size. The rate increased from 2.3 to 12.4 m/sec, when the fluence was raised from 3.18 to 11.93 J/cm2. Laser drilling wit h 193 nm light has two kinds of laser matter interaction in the drilling process, namely photochemical ablation (PCA) and photothermal ablation (PTA), which will simultaneously affect the dril ling rate of the material.226 During the ablation proc ess, the target material absorbed laser energy initially, resulting in the melting of the top layer of the SiC. The molten material continues to absorb laser energy, causing the vaporization of the drilled region.

PAGE 166

166 Following the vaporization of the material surface and the formation of saturated vapor pressure (recoil pressure),212 the local pressure was enhanced greatly, which led to extraction of molten material from the ablation zone. After pushing out the debris, the surface of s olid SiC on the bottom of via hole was again exposed to the laser beam and absorbed energy, becoming molten again. The via hole was produced by repeating above process many times. The debris generated by laser drilling was recast around the via hole, most of it just loosely attached on the SiC surface, which could be removed easily in the presence of photoresist followed by acetone r inse in ultrasonic bath. Figure 8 drilled on the SiC surface after the photoresist removal and the debris generated by laser drilling left on the SiC were not observed after this cleaning. Figure 8 4 shows the cross sectional view of circular via holes with the diameter distinct features among these via holes. The side wall of all these via holes had inclined slopes around 46 degrees. When the via hole depth became deeper, the drilled surface moved away from the created the i nclined sidewall. The debris observed inside the 10 ke the larger diameter via holes, which exhibited a flat bottom This problem was corrected by moving up the sample stage during the drilling to put the drilling on the laser focus plane. However, it was very difficult to obtain a cross sectional view of vi We have confirmed the effect of stage movement on improving the shape of the via hole

PAGE 167

167 with the rectangular via hole, which is described below. The dimension of the typical via hole for IIIV compound semiconductor devices is > 70source contact metal of the FET is less than 25fabricated under the source contact pads outside the device active area and the metal air bridge is used to connect the source contacts of the discrete devices in the multiple finger power device. If the via hole can be fabricated smaller enough and fitted directly under the source metal contact of the FET, this approach offers significant advantages such as simplifying the power device fabrication process by eliminating the metal air bridge, improving the device heat dissipation by moving the via hole closer to the gate area, reducing the die size by eliminate the front side source contact pad. With this approach, the source contacts of the discrete FETs are directly connected to the back side ground plan and no air bridge is needed. Figure 8 5 shows a photograph of three different sizes of rectangular via holes drilled on via holes and the circular 100 100 100 100 magnified and focused on the SiC substrate. Multiple exposures were used to drill the larger 100 100 100 100 as illustrated in Figure 8 6. diameter via holes, which exhibited a flat bottom, shown in F igure 8 7 B This problem

PAGE 168

168 was corrected by moving up the sam ple stage during the drilling. As shown in Figure 8 7 A crosssectional SEMs of via hole with a flat bottom and almost no inclined slope after the via hole been move back the original height and to the new location before drilling the next set of via holes. 8.4 Summary In this chapter, t he relationship between the via hole entrance diameter and laser drilling rate as well as energy density applied during processing of SiC substrates was investigated. The laser ablation approach for SiC via hole formation exhibits a very high drilling rate, accompanied with a sm ooth side wall and hole bottom, which will benefit subsequent metallization. The laser drilling approach exhibits advantages in via holes fabrication, and is an efficient and via holes showed a uniform size and spacing, which means the laser micromachining is attractive for large scale production in the processing of SiC microelectronic device manufacture.

PAGE 169

169 Figure 81. Schematic of the excimer laser drilling system. Ar F 2 Excimer Laser Beam Stop Turning Mirror 1 Turning Mirror 2 Turning Mirror 3 Mask Field Lens Variable Attenuator Process view Camera Part inspection Camera Projection Lens Linear motor sample sta ge Beam Path Sample z y x

PAGE 170

170 Figure 82. T h e d r i l l i n g r a t e o f S i C A ) Drilling rate of 4HSiC as a function of op ening size at energy density is 11.93 J/cm2. B ) Drilling rate as a function of energy 6 8 10 12 14 16 18 20 Rate ( m m/sec )70 48 24Diameter of The Via Hole (mm)10 2 4 6 8 10 12 2 4 6 8 10 12 14 Energy Density (J/cm2)Rate ( m m/sec ) (A ) ( B )

PAGE 171

171 Figure 83. Photograph of a SiC sample drilled with four different sizes of circular via

PAGE 172

172 Figure 8 4. Cross sectional view of SEM images of via holes with d i f f e r e n t diameter s. 20 m 20 m 10 m 10 m (A ) (B ) (C) (D)

PAGE 173

173 Figure 8 5. Photograph of arrays of rectangular via holes with the dimension of (column on the left) 1 (column on the right) 3

PAGE 174

174 Figure 86. Cross sectional view of SEM imagines of rectangular via holes with 90100 (From left to right) 2 0 m

PAGE 175

175 Figure 87. The cr o ssse ct i o n a l v i e w o f depth around 90 A) with and B) without stagemoving process, the latter one shows a pointed bottom. 20 m 20 m (A) (B)

PAGE 176

176 CHAPTER 9 CONCLUSIONS In Chapter 2, t he effects of source field plates on AlGaN/GaN High Electron Mobility Transistor reliability under off state stress conditions were investigated using step stress cycling. The source field plate enhanced the drain breakdown fr om 55V to 155V and the critical voltages for off state gate stress from 40V to 65V, as compared to the devices without the field plate, these results were consistent with the electric field simulation and drain IV characteristics of HEMTs with and without source field plate. Transmission electron microscopy (TEM) analysis was used to examine the degradation of the gate contacts and revealed the presence of cracking due to the inverse piezoelectric effect that appeared on both source and drain side of the ga te edges. In addition, the existence of a thin oxide layer between the Ni gate contact and AlGaN layer was apparent, and both Ni and oxygen diffused into the AlGaN donor layer. After the stepstress cycling, new threading dislocations, which provided addit ional passage for the gate leakage current, were also observed. The significant improvement of AlGaN/GaN HEMT stability by using Pt based gate metallization instead of the conventional Ni/Au was demonstrated in Chapter 3. The off state critical voltage was increased from around 4565 to >100V, and changes of the drain current, drain current on/off ratio, Schottky barrier height and reverse bias gate leakage were minimized. There was no degradation observed for the HEMTs wit h Pt/Ti/Au gate metallization. Bes ides electrical field, t he reverse bias gate leakage current and the stability between the gate metal contact and semiconductor contribute to the occurrence of a critical voltage. The better thermal stability and higher Schottky

PAGE 177

177 barrier height of Pt to GaN (1.23 eV compared to 1.09 eV for Ni) benefits such improvement. The effects of off state stress and sample temperature on the trap densities in AlGaN/GaN high electron mobility transistors (HEMTs) were studied in Chapter 4. Two different trap densities were obtained using the slope of subthreshold drain current measurements. The trap density dominated at the lower temperature range almost doubled from 1.64 1012 to 3.3 1012 /cm2eV after HEMTs reached a critical voltage for a permanent suddenincrease of the gate current during the off state drain voltage stepstress. The trap density at the higher temperature range was created by the ion bombardment during inductively coupled plasma (ICP) etching for device isolation etching, which only slightly increase d from 8. 1 012 and 9.2 1012 /cm2eV after the device stress. The trap densities were also strongly dependent on drain bias voltage. Measurements conducted at higher drain bias voltages exhibited larger trap density due to more hot electrons generated at these conditions. In Chapter 5, the passivation properties, including current collapse and rf performance, of HfO2, SiNx and Al2O3 on AlG aN/GaN HEMTs were compared. Both HfO2 and Al2O3 thin films deposited by ALD produce improvements in drainsource cu rrent due to a reduction of surface depletion effects but are less effective than SiNX for the given AlGaN/GaN epi structure. However, the oxides produce superior RF performance to SiNX and should offer advantages with respect to gate definition because of their reduced aspect ratios.

PAGE 178

178 Chapter 6 investigated the effect of buffer layer on AlGaN/GaN high electron mobility transistor reliability (HEMTs). Three different types of buffer layers were used in buffer layer beneath the two dimensional electron gas channel. The reliability of AlGaN/GaN HEMTs was improved significantly by employing the thinner and the composite buffer layer. The HEMTs with the thick GaN buffer layer showed the lowest critical voltage (Vcri) during off sta te drain step stress, which was increased by around 50 and 100% for devices with the composite AlGaN/GaN buffer layers and thinner GaN buffers, respectively. In addition, a similar trend was observed in the isolation breakdown voltage (Viso) measurements, with the highest Viso achieved based on thin GaN or composite buffer designs (600700V), while a much smaller Viso of ~200V was measured on HEMTs with the thick GaN buffer layers. Those improvements are attributed to the increasing of defect density, which consequently modifies the electric field at the drain side of the gate edge by using different buffer structures. T he radiation effects in the GaN based HEMTs, including the studies on the reliability of protonirradiated InAlN/GaN and the effects of proton energy and dose on the dc characteristics and reliability of AlGaN/GaN HEMTs were reported in Chapter 7. The reliability of InAlN/GaN high electron mobility transistors (HEMTs) was improved significantly after proton irradiation. The critical voltage ( Vcri) of off state drain step stress and the drain breakdown voltage (VBR) were increased more than 100% and 50%, respectively for the HEMTs irradiated with protons. The typical critical voltage for unirradiated devices was 45 to 55 V. By sharp contrast, no critical voltage was detected for proton irradiated HEMTs up to 100 V, which was limited by the instrument used in

PAGE 179

179 the experiment. In addition, the drain breakdown voltages were below 100 V in the reference devices and increased to above 150 V after 5 M eV proton irradiation with a dose of 51015 cm2. In this study, HEMTs were subjected with different energies of 515 MeV at a fixed dose of 5 1015 cm2, or to a different doses of 2 1011, 5 1013 or 2 1015 cm2 of protons at a fixed energy of 5 M eV. After electrical stressing, no degradation was observed for the drain or gate current voltage characteristics of the protonirradiated HEMTs. On the contrary, for the unirradiated HEMTs, there were the drain current decrease around ~12%, and the reverse bias gate leakage current increase more than two orders of magnitude as a result of electrical stressing. AlGaN/GaN HEMTs were irradiated with proton irradiation energies of 5, 10 and 15 MeV at fixed fluence of 5 1015 cm2. S ubthreshold drain leak age current and reverse gate I V decreased more than one order of magnitude for all cases due to the increase of resistivity of the HEMT channel after proton irradiation. More severe degradation with decreasing proton energy is due to the increase in linear energy transfer and corresponding increase in nonionizing energy loss with decreasing proton energy in the active region of the HEMTs. The dc characteristics as well as critical voltage of the drainvoltage electrical step stress of AlGaN/GaN high elect ron mobility transistors (HEMTs) were measured prior to and post 5 MeV proton irradiation at doses from 109 to 21014 cm2 to evaluate the feasibility of AlGaN/GaN HEMTs for space applications, which need to demonstrate radiation hardness of various irradi ations. On chip transmission line method (TLM) was used to extract contact and sheet resistances. The threshold of contact resistivity (RC) and transfer resistance (RT) degradation was at a proton dose of 5109 cm2, however,

PAGE 180

180 the threshold for sheet resist ance (RS) degradation was much higher at 21013 cm2 as compared to those for RT and RC. For the dc characteristics, minimal degradations of saturation drain current (IDSS), transconductance (gm), electron mobility, and sheet carrier concentration were obs erved for the samples irradiated with proton dose below 21013 cm2, while the reduction of these parameters were 15%, 9%, 41% and 16.6%, respectively, for the device irradiated with 21014 cm2 of protons. Drain breakdown voltage (VBR) and of critical vol tage (Vcri) unexpectedly increased 37% and 45%, respectively for the devices irradiated with 21014 cm2 of protons. Gate and drain pulse measurement were also conducted to study the trap characteristics. Both measurements showed apparent reduction of IDSS for the HEMTs irradiated at proton doses higher than 21012 cm2, which meant more traps were introduced for the implanted HEMTs. The improvements of drain breakdown voltage (VBR) and critical voltage (Vcri) were attributed to the modification of the depl etion region due to the introduction of a higher density of defects after irradiation at a higher dose. In Chapter 8, Ar/F2 based UV laser drilling ( = 193 nm) with a pulse width of ~30 nsec and a pulse frequency of 100 Hz has been used to fabricate verti cal electrical interconnects (vias) for AlGaN/GaN high electron mobility transistor (HEMTs) devices on silicon carbide (SiC) substrate. The relationship between the via hole entrance diameter and laser drilling rate as well as energy density applied during processing of GaN HEMTs on SiC substrates was demonstrated. A high yield of SiC through wafer via holes with a diameter of 1050 m without trenching or micromasking, which will benefit the subsequent metal plating and get a good electrical connection, can be achieved using an inductively coupled plasma etch under SF6/O2 + Ar plasma. The laser drilling

PAGE 181

181 approach for SiC via hole formation exhibits not only the high drilling rate (229870 m/min), but also generate a smooth side wall and hole bottom without extended defects, micro pillars. It is an efficient and economical alternative to conventional SiC via holes processing. Furthermore, the array of via holes showed a uniform size and spacing, which means the laser micromachining is attractive for large sca le production in the processing of SiC based GaN HEMTs manufacture.

PAGE 182

182 LIST OF REFERENCES 1 M. A. Khan, A. Bhattarai, J. N. Kuznia, and D. T. Olson, Appl. Phys. Lett. 63, 1214 (1993). 2 Y. F. Wu, A. Saxler, M. Moore, R. P. Smith, S. Sheppard, P. M. Chavarkar, T. Wisleder, U. K. Mishra, and P. Parikh, IEEE Electron Device Lett. 25, 117 (2004). 3 H. F. Sun, A. R. Alt, H. Benedickter, and C. R. Bolognesi, IEEE Electronics Lett. 45, 376 (2009). 4 S. J. Pearton, J. C. Zolper, R. J. S hul, and F. Ren, J. Appl. Phys. 86, 1 (1999). 5 J. W. Johnson, A. G. Baca, R. D. Briggs, R. J. Shul, C. Monier, F. Ren, S. J. Pearton, A. M. Dabiran, A. M. Wowchack, C. J. Polley, and P. P. Chow, Solid State Electron. 45, 1979 (2001). 6 H. Kim, V. Tilak, B M. Green, J. A. Smart, W. J. Schaff, J. R. Shealy, and L. F. Eastman, Phys. Stat. Sol. (A) 188, 203 (2001). 7 O. Ombacher, B. Foutz, J. Smart, J. R. Shearly, N. G. Weimann, K. Chu, M. Murphy, A. J. Sierakowski, W. J. Schaff, L. F. Eastman, R. Dimitrov, A Mitchell, and M. Stutzmann, J. Appl. Phy. 87, 334 (2000). 8 M. A. Khan, X. Hu, A. Tarakji, G. Simin, J. Yang, R. Gaska, and M. Shur, Appl. Phys. Lett. 77, 1339 (2000). 9 M.Singh, and J.Singh, J. Appl. Phys. 94, 2498 (2003). 10 F. Stengel, S. N. Mohammad and H. Morko, J. Appl. Phys. 80, 3031 (1996). 11 J. W. Chung, T. W. Kim and T. Palacios, IEEE IEDM 676 (2010). 12 J. W. Chung, W. E. Hoke, E. M. Chumbes and T. Palacios, IEEE Electron Device Lett. 31, 195 (2010). 13 O. Ambacher, J. Smart, J. R. Shealy, N G. Weimann, K. Chu, M. Murphy, W. J. Schaff, L. F. Eastman, R. Dimitrov, L. Wittmer, M. Stutzmann, W. Rieger, and J. Hilsenbeck, J. Appl. Phys. 85, 3222 (1999). 14 J. P. Ibbetson, P. T. Fini, K. D. Ness, S. P. DenBaars, J. S. Speck, and U. K. Mishra, App l. Phys. Lett. 77, 250 (2000). 15 P. Srivastava, J. Das, D. Visalli, M. V. Hove, P. E. Malinowski, D. Marcon, S. Lenci, K. Geens, K. Cheng, M. Leys, S. Decoutere, R. P. Mertens and G. Borghs, IEEE Electron Device Lett. 32, 30 (2011). 16 M. Hirose, Y. Takad a, K. Matsushita, K. Takagi and K. Tsuda, Phys. Status Solidi C 9, 369 (2012).

PAGE 183

183 17 S. Kolluri, S. Keller, S. P. DenBaars and U. K. Mishra, IEEE Electron Device Lett. 33, 44 (2012). 18 N. Adachi, Y. Tateno, S. Mizuno, A. Kawano, J. Nikaido and S. Sano, Microwave Symposium Digest, IEEE MTT S 507 (2005). 19 G. Meneghesso, G. Verzellesi, F. Danesin, F. Rampazzo, F. Zanon and A. Tazzoli, IEEE Trans. Dev. Mater. Reliab. 8, 332 (2008). 20 J. Joh, and J.A. del Alamo, IEEE Electron Device Lett. 29, 287 (2008). 21 J. A. del Alamo, and J. Joh, Microelectron. Reliab. 49, 1200 (2009). 22 J. Joh, L. Xia, and J. A. del Alamo, IEEE Int. Electron Devices Meet. 385 (2007). 23 S. Demirtas, J. Joh, J. A. del Alamo, Microelectron. Reliab. 50, 758 (2010). 24 Millman, Electronic d evices and circuits. Singapore: McGraw Hill international book company 384 (1985). 25 O. Ambacher, M. Eickhoff, A. Link, M. Hermann, M. Stutzmann, F. Bernardini, V. Fiorentini, Y. Smorchkova, J. Speck, U. Mishra, W. Schaff, V. Tilak, and L. F. Eastman, Ph ysica Status Solidi (c) 1878 (2003). 26 E. T. Yu, G. J. Sullivan, P. M. Asbeck, C. D. Wang, D. Qiao, and S. S. Lau, Applied Physics Letters 71, 2794 (1997). 27 S. Pearton, GaN and ZnO based Materials and Devices, Springer Series in Materials Science 156 ( 2012). 28 M. H. Kim, Y. G. Do, H.C. Kang, D.Y. Noh, and S.J. Park, Appl. Phys. Lett. 7, 2713 (2001). 29 E. Feltin, B. Beaumont, M. Laugt, P. de Mierry, P. Vennegu`es, H. Lahrech`e, M. Leroux, and P. Gibart, Appl. Phys. Lett. 79, 3230 (2001). 30 A. Dadgar M. Poschenrieder, J. Blasing, K. Fehse, A. Diez, and A. Krost, Appl. Phys. Lett. 80, 3670 (2002). 31 Y. Fu, and D.A. Gulino, 18(3), 965 (2000), J. Vac. Sci. Technol. A. 18, 965 (2000). 32 A. Watanabe, K. Fujimoto, M. Oda, T. Nakatsuka and A. Tamura, IEE E Int. Rel. Phys. Symp. 127 (1992). 33 B. Tedesco, C. Canali, F. Magistrali, A. Paccagnella and E. Zanoni, Microelectron. Eng. 19, 405 (1992). 34 H. Kim, V. Tilak, B. M. Green, H. Cha, J. A. Smart, J. R. Shealy and L. F. Eastman, IEEE IRPS, Orlando, 214 (2001).

PAGE 184

184 35 A. Sozza, C. Dua, E. Morvan, M. A. diFortePoisson, S. Delage, F. Rampazzo, A. Tazzoli, F. Danesin, G. Meneghesso, E. Zanoni, A. Curutchet, N. Malbert, N. Labat, B. Grimbert and J. C. De Jaeger, IEEE IDEM 593 (2005). 36 M. Meneghini, N. Ronchi, A. Stocco, G. Meneghesso, U. K. Mishra, Y. Pei and E. Zanoni, IEEE Trans Electron Dev. 58, 2296 (2011). 37 J. Joh, and J. A. del Alamo, IEDM Tech. Dig. 415 (2006). 38 S. R. Lee, D. D. Koleske, K. C. Cross, J. A. Floro and K. E. Waldrip, Appl. Phys. Lett. 85, 6164 (2004). 39 A. Sarua, H. Ji M. Kubal, M. J. Uren, T. Martin, K. J. Nash, K. P. Hilton and R. S. Balmer, Appl. Phys. Lett. 88, 103502 (2006). 40 S. Y. Park, C. Floresca, U. Chowdhury, J. L. Jimenez, C. Lee, E. Beam, P. Saunier, T. Balistreri and M. J. Kim, Microelectron. Reliab. 49, 478 (2009). 41 M. A. Khan, M. S. Shur, Q. C. Chen and J. N. Kuznia, Electronics Lett. 36, 2175 (1994). 42 S. C. Binari, K. Ikossi, J. A. Roussos, W. Kruppa, D. Park, H. B. Dietrich, D. D. Koleske, A. E. Wickenden and R. L. Henry, IEEE Trans. Electron. Dev. 48, 465 (2001). 43 P. Smorchkova, C. R. Elsass, J. P. Ibbetson, R. Vetury and B. Heying, J. Appl. Phys. 86, 4520 (1999). 44 P. Ibbetson, P. T. Fini, K. D. Ness, S. P. DenBaars, and J. S. Speck, Appl. Phys. Lett. 77, 25 0 (2000). 45 R. Vetury, N. Q. Zhang, S. Keller and U. K. Mishra, IEEE Trans. Electron. Dev. 48, 560 (2001). 46 M. Gassoumi, J. M. Bluet, C. Gaquiere, G. Guillot and H. Maaref, Microelectronics Journal, Microelectronics Journal 40, 1161 (2009). 47 G. Verzel lesi, M. Faqir, A. Chini, F. Fantini, IEEE IRPS, Montreal 732 (2009). 48 M. Faqir, G. Verzellesi, F. Fantini, F. Danesin, F. Rampazzo, G. Meneghesso, E. Zanoni, A. Cavallini, A. Castaldini, N. Labat, A. Touboul, and C. Dua, Microelectron. Reliab. 48, 1639 (2007). 49 L. Semra, A. Telia, and A. Soltani, Surf. Interface Anal. 42, 799 (2010). 50 R. Stoklas, D. Greguov, J. Novk, A. Vescan and P. Kordo, Appl. Phys. Lett. 93, 124103 (2008).

PAGE 185

185 51 E. J. Miller, X. Z. Dang, H. H. Wieder, P. M. Asbeck, and E. T. Yu G. J. Sullivan, and J. M. Redwing, J. Appl. Phys. 87, 8070 (2000). 52 W. L. Liu, Y. L. Chen, A. A. Balandin and K. L. Wang, J. Nanoelectronics and Optoelectronics 1, 258 (2006). 53 S. Xie, J. Yin, S. Zhang, B. Liu, W. Zhou and Z. Feng, SolidState Electr onics 53, 1183 (2009). 54 B. M. Green, K. K. Chu, E. M. Chumbes, J. A. Smart, J. R. Shealy and L. F. Eastman, IEEE Electron Dev. Lett. 21, 268 (2000). 55 W. Lu, V. Kumar, R. Schwindt, E. Piner, and I. Adesida, SolidState Electronics 46, 1441 (2002). 56 W. S. Tan, P. A. Houston, P. J. Parbrook, G. Hill, and R. J. Airey, J. Phys. D: Appl. Phys. 35, 595 (2002). 57 A.V. Vertiatchikh, L. F. Eastman, W. J. Schaff, and T. Prunty, Electronics Lett. 38, 388 (2002). 58 S. Arulkumaran, IWPSD 317 (2007). 59 P. Makara m, J. Joh, J. A. del Alamo, T. Palacios, and C. V. Thompson, Appl. Phys. Lett. 96, 233509 (2010). 60 U. Chowdhury, J. L. Jimenez, C. Lee, E. Beam, P. Saunier, T. Balistreri, S. Y. Park T. Lee, J. Wang, M.J. Kim, J. Joh and J.A. del Alamo, IEEE Electron Device Lett. 29, 1098 (2008). 61 P. Saunier, C. Lee, A. Balistreri, D. Dumka, J. Jimenez, H. Q. Tserng, M. Y. Kao, P. C. Chao, A. Souzis, I. Eliashevich, S. Guo, J. del Alamo, J. Joh and M. Shur, Proc. Device Res. Conf. 35 (2007). 62 E. Zanoni, G. Meneghes so, G. Verzellesi, F. Danesin, M. Meneghini, F. Rampazzo, A. Tazzoli, and F. Zanon, IEDM 381 (2007). 63 M. Faqir, G. Verzellesi, G. Meneghesso, E. Zanoni, and F. Fantini, IEEE Trans. Electron Devices 55, 1592 (2008). 64 G. Meneghesso, F. Rampazzo, P. Kordo, G. Verzellesi, and E. Zanoni, IEEE Trans. Electron Devices 53, 2932 (2006). 65 S. Singhal, J. C. Roberts, P. Rajagopal, T. Li, A. W. Hansen, R. Therrien, J. W. Johnson, I. C. Kizilyalli and K. J. Linthicum, Proc. IEEE Int. Rel. Phys. Symp. 95 (2006).

PAGE 186

186 6 6 C.Y. Chang, T. Anderson, J. Hite, L. Lu, C. F. Lo, B. H. Chu, D. J. Cheney, E. A. Douglas, B. P. Gila, F. Ren, G. D.Via, P. Whiting, R. Holzworth, K. S. Jones, S. Jang and S. J. Pearton, J. Vac. Sci. Technol. B 28, 1044 (2010). 67 E. Piner, S. Singhal, P Rajagopal, R. Therrien, J. C. Roberts, T. Li, A.W. Hanson, J.W. Johnson, I.C. Kizilyalli, and K.J. Linthicum, IEDM Tech. Dig. 411 (2006). 68 V. Kumar, G. Chen, S. Guo, and I. Adesida, IEEE Trans. Electron Devices 63, 1477 (2006). 69 Y. Hori, M. Kuzuhara, Y. Ando, and M. Mizuta, J. Appl. Phys. 87, 3483 (2000). 70 S. Karmalkar, M. S. Shur, G. Simin, and M. A. Khan, IEEE Trans. Electron Devices 52, 2534 (2005). 71 W. Saito, Y. Takada, M. Kuraguchi, K. Tsuda, I. Omura and T. Ogura, Journal of Applied Physics 43, 2239 (2004). 72 P. Rajagopal, T. Gehrke, J. C. Roberts, J. D. Brown, T. W. Weeks, E. L. Piner, and K. J. Linthicum, Mat. Res. Soc., 743 (2002). 73 W. Weeks, E. Piner, T. Gehrke, and K. Linthicum, U. S. Patent 6, 617, 060 (9 September 2003) and U. S. Patent 6, 649, 287 (18 November 2003). 74 N. A. Moser, J. K. Gillespie, G. D. Via, A. Crespo, M. J. Yannuzzi, G. H. Jessen, R. C. Fitch, B. Luo, F. Ren, B. P. Gila, A. H. Onstine, C. R. Abernathy, and S. J. Pearton, Appl. Phys. Lett. 83, 4178 (2003). 75 C. Canali, F. Castaldo, F. Fantini, D. Ogliari, L. Umena, and E. Zanoni, IEEE Electron Device Lett. 3, 185 (1986). 76 G. Meneghesso, F. Magistrali, D. Sala, M. Vanzi, C. Canali, and E. Zanoni, Microelectron. Reliab. 38, 497 (1998). 77 A. Mahajan, M. Arafa, P. Fay, C. Caneau, and I. Adesida, IEEE Electron Device Lett. 18, 284 (1997). 78 E. D. Readinger, J. A. Robinson, S. E. Mohney, and R. Therrien, Semicond. Sci. Technol. 20, 389 (2005). 79 E. Monroy, F. Calle, R. Ranchal, T. Palacios, M. Verdu, F. J. Sanchez, M. T. Montojo, M. Eickhoff, F. Omnes, Z. Bougrioua, and I. Moerman, Semicond. Sci. Technol. 17, L47 (2002). 80 J. Pedrs, R. Cuerdo, R. Lossy, N. Chaturvedi, J. Wrfl, and F. Calle, Phys. Stat. Sol. (c) 3, 1709 (2006). 81 H. Okada, S. Shikata, and H. Hayas hi, Jpn. J. Appl. Phys. 30, L558 (1991).

PAGE 187

187 82 L. Liu, T. S. Kang, D. A. Cullen, L. Zhou, J. Kim, C. Y. Chang, E. A.Douglas, Soohwan Jang, D. J. Smith, S. J. Pearton, W. J. Johnson, and F. Ren, J. Vac. Sci. Technol. B 29, 032204 (2011). 83 C. F. Lo, L. Liu, T S. Kang, R. Davies, B. P. Gila, S. J. Pearton, I. I. Kravchenko, O. Laboutin, Y. Cao, W. J. Johnson, and F. Ren, Electrochem. SolidState Lett. 14, H264 (2011). 84 J. W. Chung, J. C. Roberts, E. L. Piner, and T. Palacios, IEEE Eletron Device Lett. 29, 1196 (2008). 85 A.P. Zhang, L.B. Rowland, E.B. Kaminsky, J.B. Tucker, J.W. Kretchmer, A.F. Allen, J. Cook, and B.J. Edward, Electron. Lett. 39, 245 (2003). 86 H. Hasegawa, and M. Akazawa, J. Vac. Sci. Technol. B 27, 2048 (2008). 87 G. Meneghesso, F. Rossi, G. Salviati, M. J. Uren, E. Muoz, and E. Zanoni, Appl. Phys. Lett. 96, 263512 (2010). 88 S. Karmalkar, and D. M. Sathaiya and M. S. Shur, Appl. Phys. Lett. 82, 3679 (2003). 89 S. Karmalkar, N. Satyan, and D. M. Sathaiya, IEEE Electron Device Lett. 27, 87 ( 2006). 90 S. Huang, B. Shen, F. J. Xu, F. Lin, Zh. L. Miao, J. Song, L. Lu, L. B. Cen, L. W. Sang, Z. X. Qin, Z. J. Yang, and G. Y. Zhang, Semicond. Sci. Technol. 24, 055005 (2009). 91 A.P. Zhang, L. B. Rowland, E. B. Kaminsky, V. Tilak, J. C. Grande, J. T eetsov, A. Vertiatchikh, and L. F. Eastman, J. Eletron. Mater. 32, 388 (2003). 92 H. Kim, A. Vertiatchikh, R. M. Thompson, V. Tilak, T. R. Prunty, J. R. Shealy, and L. F. Eastman, Microelectron. Reliab. 43, 823 (2003). 93 M. Gassoumi, O. Fathallah, C. Gaqu iere, and H. Maaref, Physica B 405, 2337 (2010). 94 M. Gassoumi, M. M. Ben Salem, S. Saadaoui, B. Grimbert, J. Fontaine, C. gaquiere, and H. Maaref, Microelectron. Eng. 88, 370 (2011). 95 O. Mitrofanov, M. Manfra, and Nils Weimann, Appl. Phys. Lett. 82, 43 61 (2003). 96 W. S. Tan, P. A. Houston, P. J. Parbrook, D. A. Wood, G. Hill, and C. R. Whitehouse, Appl. Phys. Lett. 80, 3207 (2002). 97 92, 152113 (2008).

PAGE 188

188 98 D. C. Look, G. C. Farlow, P. J. Drevinsky, D. F. Bliss, and J. R. Sizelove, Appl. Phys. Lett. 83, 3525 (2003). 99 H. Hasegawa, T. Inagaki, S. Ootomo, and T. Hashizume, J. Vac. Sci. Technol. B 21, 1844 (2003). 100 S. Y. Xie, J. Y. Yin, S. Zhang, B. Liu, W. Zhou, and Z. H. Feng, SolidState Electronics 53, 1183 (2009). 101 J. W. Chung, X. Zhao, and T. Palacios, 65th Device Res. Conf. in Proc. 111 (2007). 102 R. Chu, L. Chen, N. Fichtenbaum, D. Brown, Z. Chen, S. Keller, S.P. Den Baars, and U.K. Mishra, IEEE Electron Dev. Lett. 29, 974 (2008). 103 S. Maroldt, C. Haupt, W. Pletschen, S. Muller, R. Quay, O. Ambacher, C. Schippel, and F. Schwierz, Jpn. J. Appl. Phys. 48, 04C083 (2009). 104 M. Higashiwaki, T. Mimura, and T. Matsui, Appl. Phys. Express 1 (2008). 105 E. Harvard, R. Brown, and J.R. Shealy, IEEE Trans. Electron Dev. 58, 87 (2011). 106 E.R. Heller, R. Vetury, and D.S. Green, IEEE Trans. Electron Dev. 58, 1091 (2011). 107 M. Higashiwaki, Yi. Pei, R. Chu, and U.K. Mishra, IEEE Trans. Electron Dev. 58, 1681 (2011). 108 J. S. Moon, D. Wong M. Hu P. Hashimoto M. Antcliffe C. McGuire M. Micovic, and P. Willardson, IEEE Electron Device Lett. 29, 834 (2008). 109 C. Roff, J. Benedikt, P. J. Tasker, D. J. Wallis, K. P. Hilton, J. O. Maclean, D. G. Hayes, M. J. Uren, and T. Martin, IEEE Trans. Electron Devices 56, 13 (2009). 110 D. Kim, V. Kuma J. Lee, M. Yan, A. Dabiran, A. Wowchak, P. Cho, and I. Adesida, IEEE Electron Device Lett. 30, 913 (2009). 111 A. L. Corrion, K. Shinohara, D. Regan I. Milosavljevic, P. Hashimoto, P. J. Willadsen, A. Schmitz, D. C. Wheeler, C. M. Butler, D. Brown, S. D. Burnham, and M. Micovic, IEEE Electron Device Lett. 31, 1 116 (2010). 112 X. A. Cao, A. A. Syed, and H. Piao, J. Appl. Phys. 105, 063707 (2009). 113 J. H. J. W. Lim, H. B. Lee, J. H. Lee, S.B. Hur, J. K. Ryu, K.S. Kim, T. H. Kim, S. Y. Song, W. G. Hur, S. T. Kim, and S. J. Pearton, 14, H205 (2011). Electrochem Solid State Lett. 14, H205 (2011). 114 B. Gila, F. Ren, and C.R. Abernathy, Mat. Sci. Eng: R 44, 151 (2004). 115 B.M. Green, K.K. Chu, E.M. Chumbes, J.A. Smart, J.R. Shealy, and L.F. Eastman, IEEE Electron Device Lett. 21, 268 (2000).

PAGE 189

189 116 B. P. Gila, M. Hlad, A. H. Onstine, R. Frazier, G. T. Thaler, A. Herrero, E. Lambers, C. R. Abernathy, S. J. Pearton, T. Anderson, S. Jang, F. Ren, N. Moser, R. C. Fitch, and M. Freund, Appl. Phys. Lett. 87, 163503 (2005). 117 S. Rajan, Yi Pei; Z. Cheng; S.P. DenBaars, a nd U.K. Mishra, Device Research Conference Tech Digest 131 (2008). 118 T. Palacios, A. Chakraborty, S. Heikman, S. Keller, S. P. DenBaars, and U. K. Mishra, IEEE Electron Device Lett. 27, 13 (2006). 119 B. Lu, O.I. Saadat, and T. Palacios, IEEE Electron D ev. Lett. 31, 990 (2010). 120 N. Tsurumi, H. Ueno, T. Murata, H. Ishida, Y. Uemoto, T. Ueda, K. Inoue, and T. Tanaka, IEEE Trans. Electron Devices 57, 980 (2010). 121 H. K. Lin, H. L. Yu, and F. H. Huang, Microwave Opt. Technol. Lett. 52, 1615 (2010). 122 S. Yagi, M. Shimizu, H. Okumura, H. Ohashi, Y. Yano, and N. Akutsu, Jpn. J. Appl. Phys. 46, 2309 (2007). 123 W. S. Tan, M. J. Uren, P. A. Houston, R. T. Green, R. S. Balmer, and T. Martin, IEEE Electron Dev. Lett. 27, 1 (2006). 124 F. Karouta, M. C. J. C. M. Krmer, J. J. M. Kwaspen, A. Grzegorczyk, P. Hageman, B. Hoex, W. M. M. Kessels, J. Klootwijk, E. Timmering, and M. K. Smit, ECS Trans. 16, 181 (2008). 125 Z. H. Liu, G. I. Ng, H. Zhou, S. Arulkumaran, and Y. K. T. Maung, Appl Phys. Lett. 98, 113506 (2011). 126 B. Luo, J.W. Johnson, J. Kim, R. Mehandru, F. Ren, B.P. Gila, A. Onstine, C.R .Abernathy, S.J. Pearton, A.G. Baca, R.D. Briggs, R.J. Shul, C. Monier, and J. Han, Appl. Phys. Lett. 80, 1661 (2002). 127 G. Verzellesia, G. Meneghessob, A. Chinia, E. Zanonib, and C. Canalia, Microelectron. Reliab. 45, 1585 (2005). 128 G. Meneghesso, G. Verzellesi, R. Pierobon, F. Rampazzo, A. Chini, U. K. Mishra, C. Canali, and E. Zanoni, IEEE Trans. Electron Dev. 51, 1554 (2004). 129 M. Singh, and J. Singh, J. Appl. Phys. 94, 2498 (2003). 130 Ambacher, J. Smart, J. R. Shealy, N. G. Weimann, K. Chu, M. Murphy, W. J. Schaff, L. F. Eastman, R. Dimitrov, L. Wittmer, M. Stutzmann, W. Rieger, and J. Hilsenbeck, J. Appl. Phys. 85, 3222 (1999). 131 S. A. Chevtchenko, F. Brunn er, J. Wrfl, and G. Trnkle, Phys. Status Solidi A 207, 1505 (2010).

PAGE 190

190 132 Y. Cao, T. Zimmermann, H. Xing, and D. Jena, Appl. Phys. Lett. 96, 042102 (2010). 133 S. Heikman, S. Keller, S. P. DenBaars, and U. K. Mishra, Appl. Phys. Lett. 81, 439 (2002). 134 Y C. Choi, M. Pophristic, B. Peres, H. Y. Cha, M. G. Spencer, and L. F. Eastman, Semicond. Sci. Technol. 22, 517 (2007). 135 E. B. Treidel, O. Hilt, F. Brunner, J. Wrfl, and G. Trnkle, Phys. Status Solidi C 6, 1373 (2009). 136 G. Y. Lee, H. H. Liu, and J I. Chyi, IEEE Electron Device Lett. 32, 1519 (2011). 137 S. Arulkumaran, T. Egawa, S. Matsui, and H. Ishikawa, Appl. Phys. Lett. 86, 123503 (2005). 138 A. Hinoki, J. Kikawa, T. Yamada, T. Tsuchiya, S. Kamiya,M. Kurouchi, K. Kosaka, T. Araki, A. Suzuki, and Y. Nanishi, Appl. Phys. Express 1, 011103 (2008). 139 A. Hinoki, S. Kamiya, T. Tsuchiya, T. Yamada, J. Kikawa, T. Araki, A. Suzuki, and Y. Nanishi, Phys. stat. sol. (c) 4, 2728 (2007). 140 S. Demirtas, and J. A. del Alamo, IEEE IRPS Proc. 134 (2010). 141 P. Ivo, A. Glowacki b, E. B. Treidel, R. Lossy, J. Wrfl, C. Boit, and G. Trnkle, Microelectron. Reliab. 51, 217 (2011). 142 C. Y. Chang, E. A. Douglas, J. Kim, L. Lu, C. F. Lo, B. H. Chu, D. J. Cheney, B. P. Gila, F. Ren, G. D. Via, David A. Cullen, L. Zhou, D. J. Smith, S. Jang, and S. J. Pearton, IEEE Trans. Device Mater. Rel. 11, 187 (2011). 143 N. Koide, H. Kato, M. Sassa, S. Yamasaki, K. Manabe, M. Hashimoto, H. Amano, K. Hiramatsu, and I. Akasaki, Cryst. Growth 115, 639 (1991). 144 X. Zhang, P. Kung, A. Saxler, D. Walker, T. Wang, and M. Rageghi, Acta Phys. Pol. A 88, 601 (1995). 145 A.M. Hofmann, D. Kovalev, G. Steude, B. K. Meyer, A. Hoffmann, L. Eckey, R. Heitz, T. Detchprom, H. Amano, and I. Akasaki, Phys. Rev. B 52, 16702 (1995). 146 H. M. C hen, Y. F. Chen, M. C. Lee, and M. S. Feng, Phys. Rev. B 56, 116942 (1997). 147 B. P. Keller, S. Keller, D. Kapolnek, M. Kato, H. Masui, S. Imagi, U. K. Mishra, and S. P. Denbaars, Electronics Lett. 31, 1102 (1995). 148 M. A. Reshchikov, and H. Morko, J. Appl. Phys. 97, 061301 (2005).

PAGE 191

191 149 H. Sun, A. R. Alt, H. Benedickter, C. R. Bolognesi, E. Feltin, J. F. Carlin, M. Gonschorek, N. Grandjean, T. Maier, and R. Quay, IEEE Electron Device Lett. 30, 796 (2009). 150 G. H. Jessen, J. K. Gillespie, G. D. Via, A. Crespo, D. Langley, M. E. Aumer, C. S. Ward, H. G. Henry, D. B. Thomson, and D. P. Partlow, IEEE Electron Device Lett. 28, 354 (2007). 151 P. Kordos, M. Mikulics, A. Fox, D.Gregusova, K. Cico, J. F. Carlin, N. Grandjean, J. Novak, and K. Frohlich, IEEE Electron Device Lett. 31 (2010). 152 H. Sun, A. R. Alt, H. Benedickter, E. Feltin, J. F. Carlin, M. Gonschorek, N. Grandjean, and C. R. Bolognesi, IEEE Electron Device Lett. 31, 293 (2010). 153 K. D. Chabak, M. Trejo, A. Crespo, D. E. Walker, J. Yang, R. Gask a, M. Kossler, J. K. Gillespie, G. H. Jessen, V. Trimble, and G. D. Via, IEEE Electron Device Lett. 31, 561 (2010). 154 S. Tirelli, D. Marti, H. Sun, A. R. Alt, H. Benedickter, E. L. Piner, and C. R. Bolognesi, IEEE Electron Device Lett. 31, 296 (2010). 155 F. Medjdoub, M. Alomari, J. F. Carlin, M. Gonschorek, E. Feltin, M. A. Py, N. Grandjean, and E. Kohn, IEEE Electron Device Lett. 29, 422 (2008). 156 C. Gaquire, F. Medjdoub, J. F. Carlin, S. Vandenbrouck, E. Delos, E. Feltin, N. Grandjean, and E. Kohn, IEEE/MTT S Int. Microwave Symp. Dig., 2145 (2007). 157 F. Medjdoub, J. F. Carlin, M. Gonschorek, M. A. Py, N. Grandjean, S. Vandenbrouck, C. Gaquire, J. C. Dejaeger, and E. Kohn, Electron. Lett. 42, 779 (2006). 158 F. Medjdoub, J. F. Carlin, C. Gaquire, N. Grandjean, and E. Kohn, The Open Electrical and Electronic Engineering Journal 2, 1 (2008). 159 J. S. Xue, Y. Hao, J. C. Zhang, X. W. Zhou, Z. Y. Liu, J. C. Ma, and Z. Y. Lin, Appl. Phys. Lett. 98, 113504 (2011). 160 H. Wang, J. W. Chung, X. Gao, S. Guo and T. Palacios, Phys. Status Solidi C 7, 2440 (2010). 161 J. Kuzmik, IEEE Electron Device Lett. 22, 510 (2001). 162 G. H. Jessen, R. C. Fitch Jr., J. K. Gillespie, G. Via, A. Crespo, D. Langley, D. J. Denninghoff, M. Trejo Jr., and E. R. Heller, IEEE El ectron Device Lett. 54, 2589 (2007). 163 C. F. Lo, L. Liu, H. Y. Kim, J. Kim, S. J. Pearton, O. Laboutin, Y. Cao, Wayne J. Johnson, I. I. Kravchenko, and F. Ren, J. Vac. Sci. Technol. B 29, 061201 (2011).

PAGE 192

192 164 H. Y. Kim, C. F. Lo, L. Liu, F. Ren, J. Kim, an d S. J. Pearton, Appl. Phys. Lett. 100, 012107 (2012). 165 X. W. Hu, A. P. Karmarkar, B. Jun, D. M. Fleetwood, R. D. Schrimpf, R. D. Geil, R. A. Weller, B. D. White, M. Bataiev, L. J. Brillson, and U. K. Mishra, IEEE Trans. Nucl. Sci. 50, 1791 (2003). 166 H. Y. Kim, J. H. Kim, S. P. Yun, K. R. Kim, T. J. Anderson, F. Ren, and S. J. Pearton, Journal of The Electrochemical Society 155, H513 (2008). 167 A. P. Kharmarkar, B. Jun, D. Fleetwood, R. D. Schrimpf, R. A. Weller, B. D. White, L. J. Brillson, and U.K. Mishra, IEEE Trans. Nucl. Sci. 51, 3801 (2004). 168 R. Oberhuber, G. Zandler, and P. Vogl, Appl. Phys. Lett. 73, 818 (1998). 169 B. D. White, M. Bataiev, S. H. Goss, X. Hu, A. Karmarkar, D. M. Fleetwood, R. D. Schrimpf, W. J. Schaff, and L. J. Brillson, IEEE Trans. Nucl. Sci. 50, 1934 (2003). 170 J. Antoszewski, M. Gracey, J. M. Dell, L. Faraone, T. A. Fisher, G. Parish, Y. F. Yu, and U. K. Mishra, J. Appl. Phys. 87, 3900 (2000). 171 J. H. Leach, C. Y. Zhu, M. Wu, X. Ni, X. Li, J. Xie, zgr, H. Morko, J. Liberis, E. Sermuksnis, A. Matulionis, H. Cheng, and Kurdak, Appl. Phys. Lett. 95, 223504 (2009). 172 J. Kuzmik, G. Pozzovivo, C. Ostermaier, G. Strasser, D. Pogany, E. Gornik, J. F. Carlin, M. Gonschorek, E. Feltin, and N. Grandjean, J. Appl. Phys. 106, 124503 (2009). 173 I. H. J. K. H. Cho, Y. H. Choi, J. Lim, Y. S. Kim, K. R. Kim and M. K. Han, Proceedings of the 20th International Symposium on Power Semiconductor Devices & IC's 241 (May 2008). 174 J. Elsner, R. Jones, M. I. Heggie, R. K. Sitch, M. Haugk, Th. Frauenheim, S. berg, and P. R. Briddon, Phys. Rev. B 58, 12572 (1998). 175 J. K. A. Hinoki, T. Yamada, T. Tsuchiya, S. Kamiya,M. Kurouchi, K. Kosaka, T. Araki, A. Suzuki, and Y. Nanishi, Appl. Phys. Express 1, 011103 (2008). 176 M. Osinski, P. Perlin, H. Schone, A. H. Paxton, and E. W. Taylor, Electron. Lett. 33, 1252 (1997). 177 S. J. Cai, Y. S. Tang, R. Li, Y. Y. Wei, L. Wong, Y. L. Chen, K. L. Wang, M. Chen, Y. F. Zhao, R. D. Schrimpf, J. C. Keay, and K. F. Galloway, IEEE Trans. Electro n. Dev. 47, 304 (2000). 178 A. D. White, M. Bataiev, S. H. Goss, X. Hu, A. Karmarkar, D. M. Fleetwood, R. D. Schrimpf, W. J. Schaff, and L. J. Brillson, IEEE Trans. Nucl. Sci. 50, 1934 (2003).

PAGE 193

1 93 179 H. Y. Kim, T. Anderson, M. A. Mastron, J. A. Freitas Jr., S .W. Jang, J. Hite, C. R. Eddy Jr., and J.Y. Kim, J. Cryst. Growth 62, 326 (2011). 180 B. Luo, J. Kim, F. Ren, J. K. Gillespie, and R. C. Fitch, Appl. Phys. Lett. 82, 1428 (2003). 181 G. Sonia, E. Richter, F. Brunner, A. Denker, R. Lossy, M. Mai, and F. Lenk, Solid State Electron. 52, 1011 (2008). 182 X. W. Hu, B. K. Choi, H. J. Barnaby, D. M. Fleetwood, R. D. Schrimpf, S. Lee, S. S. Ardalan, R. Wilkins, U. K. Mishra, and R. W. Dettmer, IEEE Trans. Nucl. Sci. 51, 293 (2004). 183 J. R. Srour, C. J. Marshall, and P. W. Marshall, IEEE Trans. Nucl. Sci. 50, 653 (2003). 184 G. P. Summers, E. A. Burke, P. Shapiro, S. R. Messenger, and R. J. Waters, IEEE Trans. Nucl. Sci. 40, 1372 (1993). 185 T. Roy, E. X. Zhang, Y. S. Puzyrev, D. M. Fleetwood, R. D. Schrimp, B. K. Choi, A. B. Hmelo, and S. T. Pantelides, IEEE Trans. Nucl. Sci. 57, 3060 (2010). 186 Y. S. Puzyrev, T. Roy, E. X. Zhang, D. M. Fleetwood, R. D. Schrimpf, and S. T. Pantelides, IEEE Trans. Nucl. Sci. 58, 2918 (2011). 187 E. A. Douglas, C. Y. Chang, B. P. Gi la, M. R. Holzworth, K.S. Jones, L. Liu, J. Kim, S. Jang, G. D. Via, F. Ren, and S. J. Pearton, Microelectron. Reliab. 52, 23 (2012). 188 B. Luo, J. Y. Kim, F. Ren, A. G. Baca, R. D. Briggs, B. P. Gila, A. H. Onstine, K. K. Allums, C. R. Abernathy, S. J. P earton, R. Dwivedi, T. N. Fogarty, and R. Wilkins, Electrochem. Solid St. 6, G31 (2003). 189 R. C. Fitch., D. E. Walker, Jr., K. D. Chabak, J. K. Gillespie, M. Kossler, M. Trejo, L. Liu, T. S. Kang, C.F. Lo, F. Ren, D. J. Cheney, and S. J. Pearton, J. Vac Sci. Technol. B 29, 061204 (2011). 190 H. Morkoc, Nitride Semiconductors and Devices, Springer, Berlin (1999). 191 Y. Irokawa, Y. Nakano, M. Ishiko, T. Kachi, J. Kim, F. Ren, B.P. Gila, A. Onstine, C.R. Abernathy, S.J. Pearton, C.C. Pan, G.T. Chen, and J .I. Chyi, Appl. Phys. Lett. 84, 2919 (2004). 192 M. A. Khan, X. Hu, A. Tarakji, G. Simin, J. Yang, R. Gaska, and M.S. Shur, IEEE Electron. Dev. Lett. 21, 63 (2000). 193 C. Claeys, and E. Simoen, Radiation Effects in Advanced Semiconductor Materials and Dev ices, Springer Verlag (2000).

PAGE 194

194 194 H. Y. Kim, J. Kim, L. Liu, C.F. Lo, F. Ren, and S.J. Pearton, J. Vac. Sci. Technol. B 30, 012202 (2012). 195 L. Liu, C. F. Lo, Y. Y. Xi, Y. X. Wang, F. Ren, S. J. Pearton, B. P. Gila, H.Y. Kim, J. Kim, R. C. Fitch, D. E. Walker Jr., K. D. Chabak, J. K. Gillespie, S. E. Tetlak, D. Via, A. Crespo, and I. I. Kravchenko, J. Vac. Sci. Technol. B 31, 022201 (2013). 196 X. Hu, B.K. Choi, H.J. Barnaby, D.M. Fleetwood, R.D. Schrimpf, S. Lee, S. Shojah Ardalan, R. Wilkins, U.M. Mishra, and R.W. Dettmer, IEEE Trans. Nucl. Sci. 51, 293 (2004). 197 C. F. Lo, L. Liu, H. Y. Kim, J. Kim, S. J. Pearton, O. Laboutin, Yu Cao, Wayne J. Johnson, I. I. Kravchenko, and F. Ren, J. Vac. Sci. Technol. B 29, 061201 (2011). 198 A.Y. Polaykov, S.J. P earton, P. Frenzer, F. Ren, L. Liu, and J. Kim, J. Mater. Chem. C 1, 877 (2013). 199 D.W. Jenkins, J.D. Dow, and M. H. Tsai, J. Appl. Phys. 72, 4130 (1992). 200 J. Neugebauer, and C.G. Van de Walle, Phys. Rev. B. 50, 8067 (1994). 201 P. Boguslawski, E.I. B riggs, and J. Bernholc, Phys. Rev. B 51, 17255 (1995). 202 L. Zhou, J. H. Leach, X. Ni, H. Morko, and D. J. Smith, J. Appl. Phys. 107, 014508 (2010). 203 A.P. Karmarkar, B. Jun, D. M. Fleetwood, R. D. Schrimpf, R. A.Weller, B. D. White, L. J. Brillson, an d U. K. Mishra, IEEE Trans. Nuclear Sci. 51, 3801 (2004). 204 Q. Feng, L.M. Li, Y. Hao, J. Y. Ni, and J. C. Zhang, SolidState Electron. 53, 955 (2009). 205 A.Hinoki, J. Kikawa, T. Yamada, T. Tsuchiya, S. Kamiya,M. Kurouchi, K. Kosaka, T. Araki, A. Suzuk i, and Y. Nanishi, Appl. Phys. Express 1, 011103 (2008). 206 L. Liu, C. F. Lo, Y. Y. Xi, F. Ren, S. J. Pearton, O. Laboutin, Y. Cao, J. W. Johnson, and I. I. Kravchenko, J. Vac. Sci. Technol. B 31, 011805 (2013). 207 M. Mehregany, and C. A. Zorman, Thin Solid Films 518, 355 (1999). 208 M. Ostling, S. M. Koo, C. M. Zetterling, S. Khartsev, and A. Grishin, Thin Solid Films 469, 444 (2004). 209 D. J. Young, J. G. Du, C. A. Zorman, and W. H. Ko, IEEE Sensors Journal 4, 446 (2004). 210 O. Krger, G. Schne, T. W ernicke, R. Lossy, A. Liero, F. Schnieder, J. Wrfl, and G. Trnkle, IEEE Electron Dev. Lett. 27, 425 (2006).

PAGE 195

195 211 R. Lossy, A. Liero, O. Krger, J. Wrfl, and G. Trnkle, Phys. Stat. Sol. (c) 3, 482 (2006). 212 M. Schneider, L. Berthe, R. Fabbro, and M. Muller, J. Phys. D: Appl. Phys. 41, 155502 (2008). 213 M. S. Giridhar, K. Seong, A. Schulzgen, and P. Khulbe, Appl. Opt. 43, 4584 (2004). 214 Y. Sun, Y. C. Kwok, and N. T. Nguyen, J. Micromech. Micro. Eng 16, 1681 (2006). 215 T. Gumpenberger, T. Sato, R. Kur osaki, A. Narazaki, Y. Kawaguchi, and H. Niino, Chem. Lett. 35, 218 (2006). 216 H. Cho, K. P. Lee, P. Leerungnawarat, S. N. G. Chu, F. Ren, S. J. Peartona, and C. M. Zetterling, J. Vac. Sci. Technol. A 19, 1878 (2001). 217 D. W. Kim, H. Y. Lee, B. J. Park, H.S. Kim, Y. J. Sung, S. H. Chae, Y. W. Ko, and G. Y. Yeom, Thin Solid Films 447, 100 (2004). 218 B. W. Kim, S. Y. Lee, and B. T. Lee, Microelec. Eng. 71, 329 (2004). 219 F.A. Khan, and I. Adesida, Appl. Phys. Lett. 75, 2268 (1999). 220 G. Benheim, and C. S. Salupo, Mat. Res. Soc. Symp. 622, T8.9.1 (2000). 221 P. Leerungnarawat, D.C. Hays, H. Cho, S.J. Pearton, R.M. Strong, C.M. Zetterling, and M. Ostling, J. Vac. Sci. Technol. B 17, 2050 (1999). 222 L. D. Jiang, and R. Cheung, Microelec. Eng. 73, 306 (2004). 223 S. H. Kuaha, and P. C. Wood, J. Vac. Sci. Technol. A 23, 947 (2005). 224 L. F. Voss, K. Ip, S. J. Pearton, R. J. Shul, M. E. Overberg, A. G. Baca, C. Sanchez, J. Stevens, M. Martinez, M. G. Armendariz, and G. A. Wouters, J. Vac. Sci. Technol. 26, 487 (2008). 225 A. Szczesny, P. Sniecikowski, J. Szmidt, and A. Werbowy, Vacuum 70, 249 (2003). 226 P. Chabert, N. Proust, J. Perrin, and R. W. Boswell, Appl. Phys. Lett. 76, 2310 (2000). 227 S. Tanaka, K. Rajanna, T. Abe, and M. Eshashi, J. Vac. Sci. Technol. B 19, 2173 (2001). 228 P. Chabert, J. Vac. Sci. Technol. B 19, 1339 (2001). 229 N. Okamoto, J. Vac. Sci. Technol. A 27, 295 (2009). 230 J. P. Desbiens, and P. Masson, Sensors and Actuators A 136, 554 (2007).

PAGE 196

196 231 K. H. Chen, W. H. Wub, B. H. Chu, C. Y. C hanga, J. S. Lin, S. J. Pearton, D. P. Norton, and F. Ren, Appl. Sur. Sci. 256, 183 (2009). 232 M. Eyett, and D. Bauerle, Appl. Phys. Lett. 51, 2054 (1987). 233 H. Schmidt, J. Ihlemann, B. Wolff Rottke, K. Luther, and J. Troec, J. of Appl. Phys. 83, 5458 ( 1998).

PAGE 197

197 BIOGRAPHICAL SKETCH Lu Liu was born in Beijing, China. He began his higher education at Beijing University of Technology in China and earned his Bachelor of Engineering in 2005. From 2005 to 2008, Lu Liu served as research assistant at Beij ing University of Technology and Peking University under the guidance of Dr. Hong He, Dr. Haichao Liu and Dr. Jun Ma, his major was heterocatalysis. In December 2009, Lu Liu joined Dr. Fan Rens research group at University of Florida and received a Master of Engineering in Chemical Engineering in December 2010. In spring 2011, Dr. Lu Liu enrolled in the Ph.D. program and continued the research in Dr. Fan Rens group to further explore his horizon in the field of compound semiconductor microelectronics. Since joining the group Dr. Lu Lius research has focused mainly on the development of GaN HEMTs technology, including the fabrication, characterization as well as the study on the reliability and radiation effects in GaN based HEMTs. After an intensive learning and research study, Dr. Lu Liu graduated in December 2013, with a Doctor of Philosophy in chemical engineering.


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EQS34LO01_6ALPND INGEST_TIME 2014-05-13T21:07:50Z PACKAGE UFE0046322_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES