<%BANNER%>

Fabrication and Characterization of Heterojunction Transistors

Permanent Link: http://ufdc.ufl.edu/UFE0042897/00001

Material Information

Title: Fabrication and Characterization of Heterojunction Transistors
Physical Description: 1 online resource (142 p.)
Language: english
Creator: LO,CHIEN-FONG
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: ALGAN -- ALN -- GAN -- HBT -- HEMT -- INALN -- INGAASSB -- SENSOR
Chemical Engineering -- Dissertations, Academic -- UF
Genre: Chemical Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Submircon emitter finger high-speed double heterojunction InAlAs/InGaAsSb/InGaAs bipolar transistors (DHBTs) and a variety of nitride high electron mobility transistors (HEMTs) including AlGaN/GaN, InAlN/GaN, and AlN/GaN were fabricated and characterized. DHBT structures were grown by solid source molecular beam epitaxy (SSMBE) on Fe-doped semi-insulating InP substrates and nitride HEMTs were grown with a metal organic chemical vapor deposition (MOCVD) system on sapphire or SiC substrates. AlN/GaN HEMTs were grown with a RF?MBE on sapphire substrates. Ultra low base contact resistance of 3.7 ? 10-7 ohm-cm2 after 1 min 250?C thermal treatment on noval InGaAsSb base of DHBTs was achieved and a long-term thermal stability of base metallization was studied. Regarding small scale DHBT fabrication, tri-layer system was introduced to improve the resolution for submicron emitter patterning and help to pile up a thicker emitter metal stack; guard-ring technique was applied around the emitter periphery in order to preserve the current gain at small emitter dimensions. Ultra low turn-on voltage and high current gain can be realized with InGaAsSb-base DHBTs as compared to the conventional InGaAs-base DHBTs. A peak current gain cutoff frequency (fT) of 268 GHz and power gain cutoff frequency (fmax) of 485 GHz were achieved. 15 GaN-based HEMTs herein were fabricated with gate lengths from 400 nm to 1?m, and were deposited Ti/Al/Ni/Au as their Ohmic contact metallization. Effects of the Ohmic contact annealing for lattice-matched InAlN/GaN HEMTs with and without a thin GaN cap layer were exhibited and their optimal annealing temperature were obtained. A maximum drain current of 1.3 A/mm and an extrinsic transconductance of 366 mS/mm were demonstrated for InAlN/GaN HEMTs with the shortest gate length. A unity-gain cutoff frequency (fT) of 69 GHz and a maximum frequency of oscillation (fmax) of 80 GHz for InAlN/GaN HEMTs were extracted from measured scattering parameters. Passivation is one of the most important parts in device processing for preventing degradation from various environmental conditions and promising a better device performance. Simply, ozone treatment of AlN on AlN/GaN heterostructures produced effective aluminum oxide surface passivation and chemical resistance to the AZ positive photoresist developer used for subsequent device fabrication. Metal oxide semiconductor diode-like gate current-voltage characteristics and minimal drain current degradation during gate pulse measurements were observed. With an additional oxygen plasma treatment on the gate area prior to the gate metal deposition, enhancement-mode AlN/GaN HEMTs were realized. In addition, for AlGaN/GaN HEMTs in high electrical field applications, a high-dielectric-strength SiNx passivation over an optimum thickness was needed to suppress surface flashover during a high voltage or high power operation. An excellent isolation blocking voltage of 900 V with a leakage current at 1 ?A/mm was obtained across a nitrogen-implanted isolation-gap of 10 ?m between two Ohmic pads. The radiation hardness of HBTs and HEMTs is one of the critical factors that need to be established for military, space, and nuclear industry applications. The effects of proton radiation on the dc performance of InAlAs/InGaAsSb/InGaAs HBTs and AlN/GaN HEMTs were 16 investigated. Both of these devices showed a remarkable resistance to high energy proton-induced degradation and appeared very promising for terrestrial or space-borne applications. The proton-irradiated devices with a dose of 2 ? 1011 cm-2 (estimated to be equivalent to more than 40 years of exposure in low-earth orbit) showed only small changes in dc transfer characteristics, threshold voltage shift, and gate-lag with a high frequency pulse on the gate of the HEMTs and showed small changes in junction ideality factor, generation recombination leakage current, and output conductance for the HBTs. The effect the gate metallization on the nitride HEMT reliability was also examined. By replacing the conventional Ni/Au gate metallization with Pt/Ti/Au, the critical voltage for degradation of AlGaN/GaN HEMTs during off-state biasing stress was significantly improved from -55 V to over larger than -100 V. Besides the irradiation or high voltage stresses, the effects of ambient on the Pt-gated HEMT sensor for gas sensing application were also explored. For the hydrogen sensing, the sensitivity decreased proportional to the relative humidity but the presence of humidity dramatically improved the sensor recovery characteristics after exposure to the hydrogen ambient.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by CHIEN-FONG LO.
Thesis: Thesis (Ph.D.)--University of Florida, 2011.
Local: Adviser: Ren, Fan.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0042897:00001

Permanent Link: http://ufdc.ufl.edu/UFE0042897/00001

Material Information

Title: Fabrication and Characterization of Heterojunction Transistors
Physical Description: 1 online resource (142 p.)
Language: english
Creator: LO,CHIEN-FONG
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2011

Subjects

Subjects / Keywords: ALGAN -- ALN -- GAN -- HBT -- HEMT -- INALN -- INGAASSB -- SENSOR
Chemical Engineering -- Dissertations, Academic -- UF
Genre: Chemical Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Submircon emitter finger high-speed double heterojunction InAlAs/InGaAsSb/InGaAs bipolar transistors (DHBTs) and a variety of nitride high electron mobility transistors (HEMTs) including AlGaN/GaN, InAlN/GaN, and AlN/GaN were fabricated and characterized. DHBT structures were grown by solid source molecular beam epitaxy (SSMBE) on Fe-doped semi-insulating InP substrates and nitride HEMTs were grown with a metal organic chemical vapor deposition (MOCVD) system on sapphire or SiC substrates. AlN/GaN HEMTs were grown with a RF?MBE on sapphire substrates. Ultra low base contact resistance of 3.7 ? 10-7 ohm-cm2 after 1 min 250?C thermal treatment on noval InGaAsSb base of DHBTs was achieved and a long-term thermal stability of base metallization was studied. Regarding small scale DHBT fabrication, tri-layer system was introduced to improve the resolution for submicron emitter patterning and help to pile up a thicker emitter metal stack; guard-ring technique was applied around the emitter periphery in order to preserve the current gain at small emitter dimensions. Ultra low turn-on voltage and high current gain can be realized with InGaAsSb-base DHBTs as compared to the conventional InGaAs-base DHBTs. A peak current gain cutoff frequency (fT) of 268 GHz and power gain cutoff frequency (fmax) of 485 GHz were achieved. 15 GaN-based HEMTs herein were fabricated with gate lengths from 400 nm to 1?m, and were deposited Ti/Al/Ni/Au as their Ohmic contact metallization. Effects of the Ohmic contact annealing for lattice-matched InAlN/GaN HEMTs with and without a thin GaN cap layer were exhibited and their optimal annealing temperature were obtained. A maximum drain current of 1.3 A/mm and an extrinsic transconductance of 366 mS/mm were demonstrated for InAlN/GaN HEMTs with the shortest gate length. A unity-gain cutoff frequency (fT) of 69 GHz and a maximum frequency of oscillation (fmax) of 80 GHz for InAlN/GaN HEMTs were extracted from measured scattering parameters. Passivation is one of the most important parts in device processing for preventing degradation from various environmental conditions and promising a better device performance. Simply, ozone treatment of AlN on AlN/GaN heterostructures produced effective aluminum oxide surface passivation and chemical resistance to the AZ positive photoresist developer used for subsequent device fabrication. Metal oxide semiconductor diode-like gate current-voltage characteristics and minimal drain current degradation during gate pulse measurements were observed. With an additional oxygen plasma treatment on the gate area prior to the gate metal deposition, enhancement-mode AlN/GaN HEMTs were realized. In addition, for AlGaN/GaN HEMTs in high electrical field applications, a high-dielectric-strength SiNx passivation over an optimum thickness was needed to suppress surface flashover during a high voltage or high power operation. An excellent isolation blocking voltage of 900 V with a leakage current at 1 ?A/mm was obtained across a nitrogen-implanted isolation-gap of 10 ?m between two Ohmic pads. The radiation hardness of HBTs and HEMTs is one of the critical factors that need to be established for military, space, and nuclear industry applications. The effects of proton radiation on the dc performance of InAlAs/InGaAsSb/InGaAs HBTs and AlN/GaN HEMTs were 16 investigated. Both of these devices showed a remarkable resistance to high energy proton-induced degradation and appeared very promising for terrestrial or space-borne applications. The proton-irradiated devices with a dose of 2 ? 1011 cm-2 (estimated to be equivalent to more than 40 years of exposure in low-earth orbit) showed only small changes in dc transfer characteristics, threshold voltage shift, and gate-lag with a high frequency pulse on the gate of the HEMTs and showed small changes in junction ideality factor, generation recombination leakage current, and output conductance for the HBTs. The effect the gate metallization on the nitride HEMT reliability was also examined. By replacing the conventional Ni/Au gate metallization with Pt/Ti/Au, the critical voltage for degradation of AlGaN/GaN HEMTs during off-state biasing stress was significantly improved from -55 V to over larger than -100 V. Besides the irradiation or high voltage stresses, the effects of ambient on the Pt-gated HEMT sensor for gas sensing application were also explored. For the hydrogen sensing, the sensitivity decreased proportional to the relative humidity but the presence of humidity dramatically improved the sensor recovery characteristics after exposure to the hydrogen ambient.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by CHIEN-FONG LO.
Thesis: Thesis (Ph.D.)--University of Florida, 2011.
Local: Adviser: Ren, Fan.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2011
System ID: UFE0042897:00001


This item has the following downloads:


Full Text

PAGE 1

1 FABRICATION AND CHARACTERIZATION OF HETEROJUNCTION TRANSISTORS By CHIEN FONG LO A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOC TOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 20 11

PAGE 2

2 20 11 Chien Fong Lo

PAGE 3

3 To my family with all my love

PAGE 4

4 ACKNOWLEDGMENTS The research in this dissertation comes largely from receiving great support. First of all, I would like to thank my research advisor, Prof Fan Ren, for his guidance and technical support throughout every aspect of this work. He has always guided me how to conduct a work successfully shown me his professional attitudes to research and t aught me his tremendous experience and knowledge that helped me not only to complete my degree but also to develop myself as a top notch researcher. I d eeply appreciate all the opportunities which he provided to me to collaborate with many experts in di fferent fields and to operate many precious instruments I am also indebted to the other members of my advisory committee: Prof. Steve Pearton, Prof. Jenshan Lin, and Prof. Yiider Tseng and I would like to sincerely thank Prof. Jen Inn Chyi for his instr uction and support on the HBT works Their significant contributions to this work are greatly appreciated. I am honored to have been associated with such an eminent committee. I would like to express my special thanks to all the former and current memb ers in my group for their great assistance in many ways, Dr. Jerry Wayne Johnson Dr. Jih Yun Kim Dr. Soohwan Jang Dr. Travis James Anderson Dr. Chih Yang Chang Dr. Yu Lin Wang Dr. Shengchun Hung Dr. Ke Hung Chen Dr. Byung Hwan Chu Lu Liu, Tsung Sh eng Kang, Xiao Tie Wang Rob Finch and Shao Tsu Hung I also thank my co workers and colleagues in other groups Dr. Brent Gila Dr. Young Woo Heo, David Cheney, and Erica Douglas in the Materials Science and Engineering ; Dr. Austin Ying Kuang Chen, Jaso n Te Yu Kao, and Xiaogang Yu in the Electrical and Computer Engineering; Dr. Shu Han Chen and Sheng Yu Wang in the Optics and Photonics I especially thank Dr. Brent Gila for all of the help with materials analysis, equipment service, and troubleshooting. Special thanks also go to my collaborators, Dr. Ivan Kravchenko in Oak Ridge National Laboratory for the supports on contact masks and the troubleshooting on the e beam writing; Dr. Amir Dabiran in SVT Associates, Inc.; Dr. Yu Cao

PAGE 5

5 and Dr. Oleg Laboutin i n Kopin Corp. The support staff in the Chemical Engineering has been first class: Mr. Dennis Vince and Mr. Jim Hinnant in the machine shop and Dr. Santiago Alves Tavares in the IT shop I also thank many friends from Taiwan I met here, in particular for Chih Hung Sun, Sheng Min Lin, Cheng Chun Peng, Shen Hsiu Hung, Wei Cheng, Tze Bin Song, and Chien Tsung Chen. They have been there with me to get through the bad times and share the good times in this memorable journey To my family, I am eternally indeb ted. I deeply thank my parents and my sister especially for all of their unwavering love, encouragement, and support they have given me over the years I wish there was a way for me to express the deepness of my love and appreciation. Finally I thank m y wife, Cheng Wei Hwu, for sharing everything of our li fe supporting me always on numerous aspects and bringi ng me immeasurable impact s on my life She is the love of my life, and my very best friend as well. With all of their invaluable support, this dissertation and my Ph.D. degree could be therefore accomplished. I truly appreciate it.

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ ............... 4 LIST OF TABLES ................................ ................................ ................................ ........................... 8 LIST OF FIGURES ................................ ................................ ................................ ......................... 9 ABSTRACT ................................ ................................ ................................ ................................ ... 14 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .................. 17 1.1 Motivation ................................ ................................ ................................ ......................... 17 1.1.1 Heterojunction Bipolar Transistors ................................ ................................ ........ 18 1.1.2 High Electro n Mobility Transistors ................................ ................................ ........ 21 1.2 Dissertation Outline ................................ ................................ ................................ .......... 22 2 SMALL SCALE INGAASSB BASE DOUBLE HETROJUNCTION BIPOLAR TRANSISTOR ................................ ................................ ................................ ........................ 27 2.1 Background ................................ ................................ ................................ ....................... 27 2.2 Material Growth and Device Structure ................................ ................................ ............. 28 2.3 S mall Scale InGaAsSb base Double Hetrojunction Bipolar Transistor Fabrication ........ 28 2.4 Small Scale InGaAsSb base Double Hetrojunction Bipolar Transistor Electrical Characterization ................................ ................................ ................................ .................. 37 3 EFFECTS OF OHMIC METAL ANNEALING ON INALN/GAN HIGH ELECTRON MOBILITY TRANSISTOR STRUCTURES ................................ ................................ ......... 55 3.1 Background ................................ ................................ ................................ ....................... 55 3.2 Experimental ................................ ................................ ................................ ..................... 56 3.3 Annealing Temperature Dependence of Ohmic Contact Resistance and Morphology on InAlN/GaN High Electron Mobility Transistor Struct ures ................................ ............ 57 4 PASSIVATION ON GAN BASED ELECTRONIC DEVICES ................................ ........... 69 4.1 Passivation of AlN/GaN high electron mobility transistor using ozon e treatment .......... 69 4.1.1 Background ................................ ................................ ................................ ............. 69 4.1.2 Experimental ................................ ................................ ................................ ........... 70 4.1.3 R esults and Discussions ................................ ................................ ......................... 71 4.2 Effect of Silicon Nitride Passivation on Isolation Blocking Voltage in AlGaN/GaN High Electron Mobility Transistor Structure ................................ ................................ ...... 73 4.2.1 Background ................................ ................................ ................................ ............. 73 4.2.2 Experimental ................................ ................................ ................................ ........... 74

PAGE 7

7 4.2.3 Results and Discussions ................................ ................................ ......................... 75 5 EFFECTS OF PROTON IRRADIATION ON HETEROJUNCTION TRANSISTORS ...... 88 5.1 Proton Irradiation Effects on Sb based Heterojunction Bipolar Transistors .................... 88 5.1.1 Background ................................ ................................ ................................ ............. 88 5.1.2 Experimental ................................ ................................ ................................ ........... 89 5.1.3 Results and Discussio ns ................................ ................................ ......................... 90 5.2 Proton Irradiation Effects on AlN/GaN High Electron Mobility Transistors ................... 94 5.2.1 Background ................................ ................................ ................................ ............. 94 5.2.2 Experimental ................................ ................................ ................................ ........... 95 5.2.3 Results and Discussions ................................ ................................ ......................... 97 6 NI AND PT GATED ALGAN/GAN EL ECTRON MOBILITY TRANSISTORS ........... 112 6.1 Background ................................ ................................ ................................ ..................... 112 6.2 Improvement of Off State Stress Critical Voltage by Using Pt gated AlGaN/ GaN High Electron Mobility Transistors ................................ ................................ .................. 113 6.2.1 Experimental ................................ ................................ ................................ ......... 113 6.2.2 Results and Discussions ................................ ................................ ....................... 114 7 PT GATED ALGAN/GAN HIGH ELECTRON MOBILITY TRANSISTOR BASED SENSORS ................................ ................................ ................................ ............................. 121 7.1 Background ................................ ................................ ................................ ..................... 121 7.2 Effect of Humidity on Hydrogen Sensitivity of Pt Gated AlGaN/GaN High Electron Mobility Transistor Based Sensors ................................ ................................ ................... 122 7.2.1 Experimental ................................ ................................ ................................ ......... 122 7.2.2 Results and Discussions ................................ ................................ ....................... 123 8 SUMMARY ................................ ................................ ................................ .......................... 128 8.1 InAlAs/InGaAsSb/InGaAs Double Hetrojunction Bipolar Transist ors ......................... 128 8.2 InAlN /GaN High Electron Mobility Transistors ................................ ............................ 128 8.3 Passivation on AlN/GaN and AlGaN/GaN High Electron Mobility Tra nsistors ........... 128 8.4 Proton Irradiation on InAlAs/InGaAsSb/InGaAs Double Hetrojunction Bipolar AlN/GaN Transistors and High Electron Mobility Transistors ................................ ........ 129 8.5 Critical Voltage on Pt gated AlGaN/GaN Electron Mobility Transistors ...................... 130 8.6 Pt Gated AlGaN/GaN High Electron Mobility Transistor Based Hydrogen Sensors .... 130 LIST OF REFERENCES ................................ ................................ ................................ ............. 131 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ....... 141

PAGE 8

8 LIST OF TABLES Table page 1 1 Selected material properties at 300 K relevant to electronic device applications for Si, Ge, GaAs, InP, and wide bandgap semiconductors [48 50]. ................................ ........ 24 2 1 Epitaxial layer structure parameters of fabricated InAlAs/InGaAsSb/InGaAs DHBTs. ................................ ................................ ................................ .............................. 39 2 2 Sheet, contact, and transfer resistances for In 0.52 Al 0.48 As/In 0.42 G a 0.58 As 0.77 Sb 0.23 / In 0.53 Ga 0.47 As DHBT structure. ................................ ................................ .......................... 39 3 1 Thickness of the layers estimated from the simulation of the XRR and RC measurements. ................................ ................................ ................................ .................... 60 5 1 Emitter, base and collector Sheet resistance before and after proton irradiation ............. 100 5 2 Sheet, contact and transfer resistance before and after proton irradiation ....................... 100

PAGE 9

9 LIST OF FIGURES Figure page 1 1 The figure of merit (F.O.M.) of f max versus f T of InP based HBTs and SiGe HBTs for state of the art transistor techno logies. ................................ ................................ ........ 25 1 2 Bandgap energy and lattice constant of various III V compound semiconductors at room temperature (Tien, 1988). ................................ ................................ ......................... 25 1 3 Transistors per chip for Intel CPUs from 4004 Microprocessor (1971) to Intel exponential growth as doubling transistor count every two years [ 47 ]. ............................ 26 1 4 Bandgap energy and lattice constant of several III V, II VI, and elemental semiconductors at room temperature. Note the limits of the visible portion of the electromagnetic spectrum. ................................ ................................ ................................ 26 2 1 Schematic cross sectional view of a fabricated double mesa In 0.52 Al 0.48 As/ In 0.42 Ga 0.58 As 0.77 Sb 0.23 /In 0.53 Ga 0.47 As DHBT. ................................ ................................ .... 40 2 2 The device layout of a unit cell. ................................ ................................ ......................... 41 2 3 Schematic of process sequence for In 0.52 Al 0.48 As/In 0.42 Ga 0.58 As 0.77 Sb 0.23 / In 0.53 Ga 0.47 As DHBT followed the view direction of the line a in Figure 2 2). ................ 42 2 4 Schematic of process sequence for In0.52Al0.48As/In0.42Ga0.58As0.77Sb0.23/ In0.53Ga0.47As DHBT from the front side view (followed the line b in Figure 2 2). .... 43 2 5 Schematic of process sequence for emitter contact metal definition on trilayer resist system .. ................................ ................................ ................................ ............................. 44 2 6 SEM micrographs of trilevel resist structure after CF 4 /O 2 RIE (a) after soaking in developer (b), and after UV ozone treament (c). ................................ ............................... 45 2 7 SEM micrographs of as deposited emitter metal (a and b) and emitter metal after lift off process (c). ................................ ................................ ................................ ............. 46 2 8 The top 2 emitter finger (a) and the front side of emitter fingers with different widths of 0.6, 1, 1.2 m (b). ................................ ............... 47 2 9 The process sequence of guard rings and le dge fabrication (top) and the SEM micrograph of fabricated ledge structure (bottom). ................................ ........................... 48 2 10 SEM micrographs of 1 m air bridge with photoresist after isolation wet etching (a) and after th e resist strip (b). ................................ ................................ .......................... 49 2 11 SEM micrograph of 0.65 8.65 m 2 device prior to BCB planarization. ........................ 50

PAGE 10

10 2 12 SEM micrographs of exp osed emitter finger (a) and collector/base contact poles (b) after BCB etch back process. ................................ ................................ ............................. 50 2 13 The top view of optical microscope image for a fabricated device after Ti/Au interconnect m etal deposition. ................................ ................................ ........................... 51 2 14 Common emitter characteristics of the In 0.52 Al 0.48 As/In 0.42 Ga 0.58 As 0.77 Sb 0.23 / In 0.53 Ga 0.47 As DHBT. ................................ ................................ ................................ ......... 51 2 15 0.52 Al 0.48 As/In 0.42 Ga 0.58 As 0.77 Sb 0.23 / In 0.53 Ga 0.47 current (bottom). ................................ ................................ ................................ ................ 52 2 16 Current gian h 21 U rf extrapolations of 0.65 8.65 m 2 DHBT. ................................ ................................ ................................ ................................ 53 2 17 Dependence of f T and f max on collector current density J C for the In 0.52 Al 0.48 As/ In 0. 42 Ga 0.58 As 0.77 Sb 0.23 /In 0.53 Ga 0.47 As DHBT with a 0.65 8.65 m 2 emitter area. ........... 54 2 18 Overview of f T and f max data for high speed InP and Si Ge based HBTs The figure of merit (F. O.M) was defined as f T f max / ( f T + f max ). ................................ ....................... 54 3 1 Schematic of the two InAlN/GaN layer structures investigated, ie. with (a) and without (b) a thin GaN cap. ................................ ................................ ................................ 60 3 2 Variation of mobility (a) and sheet carrier density (b) as a function of annealing temperature for structures with and without a GaN cap layer. ................................ .......... 61 3 3 C ontact resistivity (a), sheet resistance (b) and transfer resistance (c) for contacts on both types of InAlN/GaN structures as a function of annealing temperature. ................... 62 3 4 Optical micrographs of contacts on InAlN/GaN structures without the GaN cap, as a function of annealing temperature. ................................ ................................ .................... 63 3 5 Optical micrographs of contacts on InAlN/GaN st ructures with the GaN cap, as a functio n of annealing temperature. ................................ ................................ .................... 64 3 6 (a) XRR curves acquired from an as grown sample and from a sample that was annealed at 850 o C; (b) High resolution omega 2theta rocking curves acquired near the (002) GaN diffraction line for control (blue trace) and annealed sample (green trace); the red trace is the simulated curve for the control sample. ................................ ... 65 3 7 Drain I V characteristics at ze ro gate voltage f or InAlN /GaN HE MTs without (a) or with (b) GaN cap layers, as a function of contact annealing temperature. ........................ 66 3 8 (a) Comparison of drain I V characteristics at 4 V gate vol tage for InAlN /GaN HEMTs with or without GaN cap layers, for the optimized contact annealing

PAGE 11

11 of gate voltage. ................................ ................................ ................................ ................... 67 3 9 Log plot of forward and reverse I V characteristics from HEMTs with and without GaN caps. The barrier heights were extracted from the forward I V characteristics. Results from two devices with GaN caps are shown to give an idea of the variation seen from device to device. ................................ ................................ ............................... 68 4 1 Schematic cross section view of the AlN/GaN HEMT grown on sapphire. ...................... 79 4 2 EDS spectra of the AlN/GaN HEMT su rface before and after the ozone treatment as well as the EDS spectrum of the AlN/GaN HEMT surface with an additional oxygen plasma exposure. ................................ ................................ ................................ ................ 80 4 3 Gate current of the AlN/GaN HEMTs treated with ozone (d mode HEMT) only and with an additional oxygen plasma exposure (e mode HEMT). ................................ ......... 80 4 4 DC IDS VDS characteristics of an AlN/GaN HEMT with (left) no oxygen plasma exposure, (right) the gate area of the HEMT exposed to 24 sec of oxygen plasma. ......... 81 4 5 DC transfer characteristics for the d mode and e mode AlN/GaN HEMTs. ..................... 81 4 6 Gate lag measurement data of the drain current as a function of pulse gate voltage for the e mode and d mode HEMTs. ................................ ................................ ................. 82 4 7 Schematic cross sectional view of an isolation blo cking voltage tester fabricated on the AlGaN/GaN HEMT structure. ................................ ................................ ..................... 83 4 8 Isolation current voltage characteristics measured across two 100 m 100 m Ohmic contact pads separated with an isol ation implanted space ranging from 1.7 or 4.7 m. These testers were passivated with different thicknesses of SiNx passivation layer. The distance, d, between the dielectric openings was kept at 140 m. .................. 84 4 9 Microscope pictures of device before (top) and after (bottom) experiencing early breakdown. ................................ ................................ ................................ ......................... 84 4 10 Comparison between ATLAS simulated isolation blocking voltages and the experimental data, where the isolation blocking voltages were plotted as the functions of the thickness of the SiN x passivation layer and the spacing of the isolation implanted region. ................................ ................................ ................................ 85 4 11 Isolation blocking voltage as a function of the distance between the two contact window openings on the samples passivated with 375 nm of SiN x layer. ......................... 85 4 12 Isolation blocking voltage as a function of the distance between the two contact window openings on the samples passivated with 375 nm or 1.125 m of SiN x layer. The samples exhibited with an early breakdown voltage were labeled with stars and triangles. ................................ ................................ ................................ ............................. 86

PAGE 12

12 4 13 I V characteristics of testers passivated with 375 nm of SiN x layer deposited at different rf powers. The separation between the contact window openings was kept at 140 m and the implanted isolation gap was 4.7 m. ................................ ................... 86 4 14 Drain I V characteristics of the HEMT. The device exhibited a saturation current >310 mA/mm and a drain breakdown voltage of 1000 V. ................................ ................ 87 5 1 Schematic of a double mesa In 0.52 Al 0.48 As/In 0.39 Ga 0.61 As 0.77 Sb 0.23 /In 0.53 Ga 0.47 As DHBT ................................ ................................ ................................ ............................ 101 5 2 Optical microscope image of the double mesa DHBT (a) before and (b) after the proton irradiation. ................................ ................................ ................................ ............ 101 5 3 E B junction forward I V characteristics (top) and ideality factor (bottom) as a function of the proton dose. ................................ ................................ ............................. 102 5 4 B C junction I V characteristics (top) and ideality factor (bottom) as a function of the proton dose. ................................ ................................ ................................ ................ 103 5 5 I V characteristics of E B, B C, and E C j unctions of a reference DHBT (top) and DHBT irradiated with the fluence of 2 10 15 cm 2 (bottom). ................................ .......... 104 5 6 (top) Gummel plots before the proton irradiation and after different proton dose ir radiation. (bottom) DC current gains as a function of the collector currents. ............... 105 5 7 Common emitter characteristics from the DHBTs before and after different proton dose irradiation. ................................ ................................ ................................ ................ 106 5 8 Cross sectional TEM of AlN/GaN HEMT structure. ................................ ...................... 107 5 9 (top) Schematic cross sectional diagram of the AlN/GaN HEMT device and ( bottom) optical microscope images of (a) pre and (b) post irradiated HEMT devices. ............... 108 5 10 (left) I DS V DS characteristics of AlN/GaN HEMTs before and after 2 10 11 protons/cm 2 irradiation (right) I DS V DS characteristics of AlN/GaN HEMTs exposed to 2 10 13 and 2 10 15 protons/cm 2 irradiation. ................................ ............................. 109 5 11 Gate current of AlN/GaN HEMTs before and after proton irradiation, and a fter subsequent thermal annealing. ................................ ................................ ......................... 109 5 12 (top) DC transfer characteristics for pre and post proton irradiated AlN/GaN HEMT devices, and (bottom) extracted threshold voltage and extrinsi c maximum transconductance as a function of proton fluence. (2 10 11 2 10 15 protons/cm 2 ) ....... 110 5 13 Gate lag measurement data for the drain current as a function of the pulsed gate voltag e ................................ ................................ ................................ ............................ 111

PAGE 13

13 6 1 Off state gate currents as a function of gate for the HEMTs fabricated with Ni/Au or PT/Ti/Au gate metallization. ................................ ................................ ............................ 117 6 2 Forward (top) and reverse (bottom) Schottky gate characteristics before and after the off state stress for the HEMTs fabricated with Ni/Au and Pt/Ti/Au gate metallization. ................................ ................................ ................................ .................... 118 6 3 Drai n I V characteristics of HEMTs before and after the off state stress for the HEMTs fabricated with Ni/Au and Pt/Ti/Au gate metallization. ................................ .... 119 6 4 XPS spectra of (a) Pt 4f XPS spectra for as d eposit Pt/GaN sample. (b) Pt 4f XPS spectra for the Pt/GaN sample annealed at 300C for 30 mins. (c) Ni 2p XPS spectra for as deposit Ni/GaN sample. (d) Ni 2p XPS spectra for the Ni/GaN sample annealed at 300C for 30 mins. (e) O 1s XPS spectra for as d eposit Ni/GaN sample. (f) O 1s XPS spectra for the Ni/GaN sample annealed at 300C for 30 mins. ............... 120 7 1 Schematic of the hydrogen sensing system. ................................ ................................ .... 125 7 2 Sensing sensitivity, defined as the ratio of diode current change to the diode current in 1% hydrogen balance with air as a function of the AlGaN/GaN HEMT diode bias voltage ................................ ................................ ................................ ............................ 125 7 3 (top) Time dependence of absolute value of the AlGaN/GaN HEMT diode current biased at 1.5 V as the gas ambient switched back and forth between 1% hydrogen balance and air with different humidity. (bottom) Hydrogen sensitivity as a function of the humidity. ................................ ................................ ................................ ................ 126 7 4 Time dependence of recovery time characteristics of the AlGaN/GaN HEMT diode current biased at 1.5 V as the gas ambient switched back and forth between 1% hydrogen balance and air with 100% humidity as well as dry air. ................................ .. 127

PAGE 14

14 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy FABRICATION AND CHARACTERIZATION OF HETEROJUNCTION TRANSISTORS By Chien Fong Lo May 2011 Chair: Fan Ren Major: Chemical Engineering Submircon emitter finger high speed double heterojunction InAlAs/InGaAsSb/InGaAs bipolar transistors (DHBTs) and a variety of nitr i de high electron mobility transistors (HEMTs) including AlGaN/GaN, InAlN/GaN, and AlN/GaN were fabricated and characterized. DHBT structures were grown by solid source molecular beam epitaxy (SSMBE) on Fe doped s emi insulating InP substrates and nitride HEMTs were grown with a metal organic chemical vapor deposition (MOCVD) system on sapphire or SiC substrates. AlN/GaN HEMTs were grown with a RF MBE on sapphire substrates. Ultra low base contact resistance of 3. 7 10 7 ohm c m 2 after 1 min 250 C thermal treatment on noval InGaAsSb base of DHBTs was achieved and a long term t hermal stability of base metallization was studied. Regarding small scale DHBT fabrication, tri layer system was introduced to improve the r esolution for submicron emitter patterning and help to pile up a thicker emitter metal stack; guard ring technique was applied around the emitter periphery in order to preserve the current gain at small emitter dimensions. Ultra low turn on voltage and hi gh current gain can be realized with InGaAsSb base D HBTs as compared to the conventional InGaAs base D HBTs. A peak current gain cutoff frequency ( f T ) of 2 68 GHz and power gain cutoff frequency ( f max ) of 485 GHz were achieved.

PAGE 15

15 GaN based HEMTs herein we and were deposited Ti/Al/Ni/Au as their Ohmic contact metallization E ffects of the Ohmic contact annealing for lattice matched InAlN/GaN HEMTs with and without a thin GaN cap layer were exhibited and th eir optimal annealing temperature were obtained. A maximum drain current of 1.3 A/mm and an extrinsic transconductance of 366 mS/mm were demonstrated for InAlN/GaN HEMTs with the shortest gate length. A unity gain cutoff frequency ( f T ) of 69 GHz and a ma ximum frequency of oscillation ( f max ) of 80 GHz for InAlN/GaN HEMTs were extracted from measured scattering parameters. Passivation is one of the most important parts in device processing for preventing degradation from various environmental conditions and promising a better device performance. Simply, o zone treatment of AlN on AlN/GaN heterostructures produce d effective aluminum oxide surface passivation and chemical resistance to the AZ positive photoresist developer used for subsequent device fabrica tion. Metal oxide semiconductor diode like gate current voltage characteristics and minimal drain current degradation during gate pulse measurements were observed. With an additional oxygen plasma treatment on the gate area prior to the gate metal deposi tion, enhancement mode AlN/GaN HEMT s were realized. In addition, f or AlGaN/GaN HEMTs in high electrical field applications, a high dielectric strength SiN x passivation over an optim um thick ness wa s needed to suppress surface flashover during a high voltag e or high power operation. An excellent isolation blocking voltage of 900 V with a leakage current at 1 A/mm was obtained across a nitrogen implanted isolation gap of 10 m between two Ohmic pads. The radiation hardness of HBTs and HEMTs is one of the c ritical factors that need to be established for military, space, and nuclear industry applications. The effects of proton radiation on the dc performance of InAlAs/InGaAsSb/InGaAs HBTs and AlN/GaN HEMTs were

PAGE 16

16 investigated. Both of these devices showed a r emarkable resistance to high energy proton induced degradation and appeared very promising for terrestrial or space borne applications. The proton irradiated devices with a dose of 2 10 11 cm 2 (estimated to be equivalent to more than 40 years of exposur e in low earth orbit) showed only small changes in dc transfer characteristics, threshold voltage shift, and gate lag with a high frequency pulse on the gate of the HEMTs and showed small changes in junction ideality factor, generation recombination leakag e current, and output conductance for the HBTs. The effect the gate metallization on the nitride HEMT reliability was also examined. By replacing the conventional Ni/Au gate metallization with Pt/Ti/Au, the critical voltage for degradation of AlGaN/GaN H EMTs during off state biasing stress was significantly improved from 55 V to over larger than 100 V. Besides the irradiation or high voltage stresses, the effects of ambient on the Pt gated HEMT sensor for gas sensing application were also explored. For the hydrogen sensing, the sensitivity decreased proportional to the relative humidity but the presence of humidity dramatically improved the sensor recovery characteristics after exposure to the hydrogen ambient.

PAGE 17

17 CHAPTER 1 INTRODUCTION 1.1 Motivatio n THz waves hold an enormous technical potential as carrier waves for wireless networks. The demand for bandwidth in wireless short range communication systems has doubled every 18 months over the last ten years. Wireless data rates over 100 gigabits per second (Gbps) will be needed in 10 years from now. Some of the applications which will require tremendous bandwidth can already be foreseen and others will emerge as technology evolves. InP based Heterojunction bipolar transistors ( HBTs ) with high elect ron mobility in base layer have demonstrated great success in ultra high speed radio frequency (rf) applications and have been currently considered as the most promising technologies for achieving terahertz (THz). For the past few years, several groups ha ve used Sb base d ternary /quaternary HBTs and field effect transistors (FETs) in combination with advanced submicron device fabrication techniques to demonstrate unit gain cut off frequencies above 0.6 THz. By extrapolation of expected future parasitic res istance and capacitance reductions, devices with a performance of 1.2 THz have been predicted. The Sb based ternary /quaternary HBTs have a tremendous potential to come with significant impact in the THz technology; the generation of THz waves and THz elec tronics. Moreover, h eterojunction bipolar transistors and high electron mobility transistors (HEMTs) are the most mature of a new generation of the Group III V compound semiconductor transistors opera ting by the use of the heteroju nctions. Wide bandgap Ga N based HEMTs which were received the most attention, have come a long way as excellently potential devices in microwave and millimeter wave application, and high temperature, high power, high breakdown field operation. To further improve DC characterist ics, large power and small rf signal

PAGE 18

18 performance of these heterojunction transistors, high quality materials with low lattice stress and less defects in the epi layers were required, as well as proper Ohmic metallization with a minimum specific contact re sistivity ( c ), electronically reliable and thermally stable gate metallization, and effective and reliable surface passivation F or practical operation in harsh environments (e.g., temperature, humidity, chemical, radiation) for commercial and military a pplication, III V electronic devices have also emerged in the past twenty years as promising candidates, due to the thermal and chemical stabilities, and irradiation tolerance of the III V materials. 1.1.1 Heterojunction Bipolar Transistors The first bipol ar junction transistor (BJT), or triode, was demo n strated by John Bardeen and Walter Brattain at the Bell Telephone Laboratories in 1947 [1,2], and the concept of heterojunction bipolar transistors (HBTs) was firstly introduced by William Shockley in 1948. Herbert Kroemer at the Radio Corporation of America (RCA) Laboratories developed a detailed theory related to HBTs and proposed to improve the emitter efficiency in 1957 [3]. Heterostructure design has been recognized as the one with tremendous potentia l advantages, compared with conventional homostructure design. However, the transistor build up was highly limited by the material growth technology before 1970s. The new situation had been coming with the liquid phase epitaxy (LPE) technology for the Gr oup III V compound semiconductor heterostructures in the early 1970s, and with two more promising technologies, metal organic chemical vapor deposition (MOCVD) and molecular beam epitaxy (MBE) in the mid 1970s. As the 1980s later, g allium arsenide (GaAs) and indium phosphide (InP) based HBTs ha ve been prosperous ly investigated [4 7]. Numerous advantages of GaAs bas ed HBTs in collector efficiency power density, and peak cut off frequency have been reported, compared with GaAs based HEMTs and MESFETs

PAGE 19

19 [ 8 ]. Furthermore, InP based HBTs exhibit much greater advantages than GaAs based HBTs in high speed and low power applications, due to its superior properties, such as lower surface recombination, smaller bandgap for reducing the turn on voltage and power d issipation, and excellent specific contact resistance with non annealed O hmic contact to InGaAs. With mature growth technologies, InP based HBTs c ould be grown in various combinations of epitaxial InP, InGaAs, and InAlAs lattice matched to InP with high c rystalline perfection and purity. The emitter materials InAlAs and InP with wide bandgap have shown the improvement of emitter efficiency. Lattice matched InGaAs and GaAsSb with narrow bandgap and higher mobility are widely used as the base materials. G enerally speaking, InP/InGaAs and InP/GaAsSb heterostructures behav e as type I and type II in band alignment, respectively. However, the electron pile u p issue suppress ing the electron transport with large tunneling recombination at the type II InP/GaAsSb emitter/base (E/B) junction limits the current gain, particularly at low current region [ 9 ] Therefore, in order to s o l ve this issue and achieve the lowest turn on voltage while existing a large conduction band discontinuity ( type I ), a compositionally g raded base emitter junction is necessarily applied to reduce electron transport barrier between InGaAs or GaAsSb base and InAlAs or InP emitter. For applications requiring higher base collector breakdown voltages and lower output conductance, a wide bandga p (InP) collector is the material of choice. For the base/collector (B/C) structure, the staggered band alignment (type II) at the B/C junction has been considered as a non collector current blocking design to not only improve a breakdown voltage but also benefit a high er current operation capability [ 10 ]. H igh peak cut off frequenc ies ( f T ) of over than 60 0 GHz at room temperature with graded InGaAs and GaAsSb base were reported by Hafez et al. and Liu et al., respectively [11,12]. In addition, a f T of 6 70 GHz and f T BV CEO product of

PAGE 20

20 2. 14 THz V with a carbon doped GaAsSb InGaAsSb graded base were proposed by Snodgrass [ 1 3 ]. Figure 1 1 showed the f T and f max of InP based and SiGe HBTs from literature [ 1 1 34 ]. The figure of merit (F.O.M.) for D type Fli p Flop ICs is defined as follows [ 20 25 ]: In the world record, the highest f T of 765 GHz at room temperature and 845 GHz at 55 InP/InGaAs p seudo morphic heterojunction bipolar transistor (PHBT) with a ultra small scaled 12.5 nm base and 55 nm collector were published in 2006 [ 22 ]. Figure 1 2 shows the variation of the energy band gap with corresponding lattice constant for the Group III V compound semiconductor s. The ternary materials of InGaAs or GaAsSb are both widely used in InP based HBTs, and the quaternary InGaAsSb material could be another promising candidate for the base layer of high speed InP based HBTs with the f ollowing reasons. First of all, t he intrinsic carrier concentration ( n i ) can be increased through narrow ing the bandgap [ 35,36 ]. Higher carrier concentrations were observed in most of the InGaAsSb layers because of the smaller bandgap of the InGaAsSb mat erial than that of the lattice matched InGaAs or GaAsSb. Therefore, the InGaAsSb base layer was suitab ly strained for a lower R sh purpose by modifying the mole fractions of In and Sb Furthermore, as compared with GaAsSb base HBTs, l ower turn on voltage, higher current gain ( ), and higher cut off frequency can be achieved with the InGaAsSb base HBTs due to the reduction of conduction band offset and the increase of valance band offset at the InP/InGaAsSb E/B junction. sh ratio and current density c an be obtained A peak f T of 238 GHz and a high current gain of 125 with a low turn on voltage of 0.35 V for InP / In 0 37 Ga 0 63 As 0 89 Sb 0 11 / In 0 53 Ga 0 47 As DHBTs have been proposed by S. H. Chen et al. in 2008 [15]

PAGE 21

21 Herein, a novel InP based DHBT with an In 0 5 2 Al 0 48 As emitter, an In 0 42 Ga 0 58 As 0 7 7 Sb 0 23 base, and an In 0 53 Ga 0 47 As collector, forming a favorable type I E/B junction and a type II B/C junction, is experimentally investigated in terms of both direct current ( dc ) and radio frequency (rf) cha racteristics. Instead of the InP emitter, the InAlAs/InGaAsSb E/B junction constitutes a higher conduction band offset, which provides a higher initial injection energy for the hot electrons across the base layer and thus, reduces the base transit time a nd increase the cut off frequency in the small signal operation [ 37 ]. W ith a type I conduction band lineup at In 0.52 Al 0.4 8 As /InGaAsSb E/B junction, the electron pile up issue can be eliminated to facilitate the electron transport across the E/B junction without a tunneling current as compared to the type II InP/GaAsSb E/B junction [ 38 ]. I n addition, with a type II conduction band lineup at InGaAsSb/InGaAs B/ C junction the current blocking effect is alleviate d as well as the Kirk effect is postponed in high current density and high speed operation simultaneously 1.1.2 High Electron Mobility Transistors As far from the mid 20th century, the silicon based semiconductor electronics industry has e 1 3). While silicon devices are dominant in the solid state market, the emerging development of non silicon devices operating at high voltage solid state switching, high power, high frequency, high temperature (>200 C), or harsh (e.g., high humidity, ch emical, radiation) environments for many military and aerospace applications is needed, due to the limitations of silicon properties, such as narrow indirect band gap, and higher intrinsic carrier concentration ( n i ) at the same temperature. Nowadays, the Group III V compound semiconductors are extremely attracted as alternative materials with their unique properties (shown in Table 1 1). GaAs and InP have their merits on ultra high speed applications due to their much higher mobility than Silicon, and InP based

PAGE 22

22 transistors have held the record of frequency response ( f T and f max ) approaching to the terahertz regime. III nitride materials with wide bandgap have been considered promising materials for high temperature, high power, high breakdown field, and o ptoelectronic applications, such as AlN, GaN, and InN with their direct bandgaps of 3.4, 6.2 and 1.9 eV, respectively. The variation of the energy band gap with corresponding lattice constant for the Group III V, II VI, and elemental semiconductors was sh own in Figure 1 4. GaN based HEMTs have been recognized as excellent candidates for high voltage, high radio frequency, and microwave or millimetre wave power applications because of their high power and high speed handling capabilities. AlGaN/GaN HEMTs h ave demonstrated larger power density ( >30 W/mm ) and efficiency with field plates on SiC in 2004 [ 39 ] and a high cutoff frequencies ( f T ) of 110 GHz on HR Si in 2009 [ 40 ] For AlN/GaN HEMTs, a large band gap of AlN (6.2 eV) provides an even better carrier confinement and lowers the gate leakage current, and the absence of alloy disorder (compared to AlGaN barriers) results in the improvement of both low and high field carrier transport. A number of variations in AlN / GaN HEMTs have been explored with notab le success in developing depletion mode ( d mode) and enhancement mode ( e mode) devices [ 41 45 ]. Moreover, lattice matched InAlN/GaN HEMTs at an In mole fraction of 0.17 have shown tremendous potential as the materials system with high thermal and chemical stability. Promising dc and output power performance has been reported for InAlN/GaN HEMTs on Si, sapphire and SiC substrates, and a record f T of 144 GHz has also been proposed in 2010 [ 46 ]. 1.2 Dissertation Outline This work deals exclusively with the fabrication and characterization of III N electronic devices, including the InP base DHBTs and GaN base HEMTs.

PAGE 23

23 D etails of device fabrication for small scaled InGaAsSb base DHBTs on InP substrates will be followed in Chapter 2. DC and small radio frequen cy signal characteristics are also given. Chapter 3 discusses the effect of Ohmic contact annealing temperature on the stability of electrical properties, surface and interface morphologies, and subsequent device performance of InAlN/GaN HEMTs. Chapter 4 investigates the passivation effects on two kinds of electronic devices. One is a thin aluminum oxide film on the AlN/GaN HEMTs for protecting AlN and acting as a gate insulator; the other is a silicon nitride layer on the AlGaN/GaN HEMT structure to com pare the effects of the isolation blocking voltage with differen t passivation conditions. Chapter 5 presents the proton irradiation effects on t wo different electronic devices, including InGaAsSb base DHBTs and AlN/GaN HEMTs In Chapter 6 off state stre ss critical voltages, as well as DC characteristics, are compared by the use of Ni/Au and Pt/Ti/Au gate metallization on AlGaN/GaN HEMTs. And then, Chapter 7 reports the electrical characteristics of AlGaN/GaN HEMT based Schottky diode exposed to the hydr ogen containing air ambient with and without relative humidity (RH) under reverse biased conditions, and also proposes a sensing mechanism to explain the decreasing sensitivity under increasing RH. Chapter 8 gives brief conclusions for all of my work.

PAGE 24

24 Ta ble 1 1. Selected material properties at 300 K relevant to electronic device applications for Si, Ge, GaAs, InP, and wide bandgap semiconductors [48 50]. Property Si Ge Dia mond 4H SiC GaAs InP InN GaN (AlGaN/GaN) AlN Energy Bandgap, E g (eV) 1.12 0.66 5.45 3.25 1.42 1.34 1.89 3.4 6.2 Saturated Electron Velocity, v sat ( 10 7 cm/s) 1.0 0.6 2.7 2.0 1.0 1.0 2.5 2.5(2.7) 1.4 Peak Electron Velocity, v max ( 10 7 cm/s) ----2.1 3.02 4.3 3.1 1.7 Breakdown Field, E B (MV/cm) 0.3 0.1 10 3 0.6 0.5 -3 1.8 Electron Mobility, n (cm 2 s) 1400 3900 2200 800 8500 5400 3200 900 (2000) 300 Hole Mobility, p (cm 2 s) 500 1900 1600 50 400 150 -50 14 Static Dielectric s 11.8 16.2 5.5 9.7 12.8 12.5 15.3 9.5 8.5 Thermal Conductivity, K) 1.5 0.58 20 30 4.9 0.5 5 0.68 0.45 1.3 2.85

PAGE 25

25 Figure 1 1. The figure of merit (F.O.M.) of f max versus f T of InP based HBTs and SiGe HBTs for state of the art transistor technologies. Figure 1 2. Bandgap energy and lattice constant of various III V compo und semiconductor s at room temperature (Tien, 1988).

PAGE 26

26 Figure 1 3. Transistors per chip for Intel CPUs from 4004 Microprocessor (1971) to Intel exponential growth as doubl ing transistor count every two years [ 47 ]. Figure 1 4. Bandgap energy and lattice constant of several III V, II VI, and elemental semiconductors at room temperature. Note the limits of the visible portion of the electromagnetic spectrum.

PAGE 27

27 C HAPTER 2 SMALL SCALE INGAASSB BASE DOUBLE HETROJUNCTION BIPOLAR TRANSISTOR 2.1 Background Quaternary InGaAsSb based double heterojunction bipolar transistors (DHBTs) have attracted a great deal of attention because of their high speed and ultra low turn on voltage compared to that of the conventional In AlAs /In 0.53 Ga 0.47 As single heterojunction bipolar transistor s [1 3 17 ] These characteristics were due to their smaller base band gap and favorable type I emitter base ( E B) junction and type II base collec tor (B C) junction T he higher valence band offset at the type I In 0.52 Al 0.48 As/In 0. 42 Ga 0. 58 As 0.7 7 Sb 0. 23 E/B junction prevented the injection of holes from the base to the In 0.52 Al 0.48 As emitter, leading to higher electron injection efficiency. Furthermo re, the type II staggered band alignment of the In 0. 42 Ga 0. 58 As 0.7 7 Sb 0. 23 /In 0.53 Ga 0.47 As ( B/C ) junction increased the current operation capability due to the postponed Kirk effect. Excellent dc performance of DHBTs with lattice matched or strained InGaAsSb as the base layer ha s been demonstrated [17,177] To further achieve high speed operation on InGaAsSb DHBTs, it is essential not only to decrease the carrier transit time regarding the base and collector layers through the structure design and wafer grow th quality, but also to reduce both the emitter size and the parasitic resistance and capacitance through the layout design and device fabrication, simultaneously. Therefore, it is desirable to create submicron emitter fingers with high resolution trileve l resist system and minimize the base resistance and capacitance by employing self aligned processing in which the device features. In this study, we describe processing sequences for In 0.52 Al 0.48 As/In 0. 42 Ga 0. 58 As 0.7 7 Sb 0. 23 / In 0.53 Ga 0.47 As DHBTs and show the examine d device dc and rf performance. The formation of the trilevel stack using polydimethylglutarimide (PMGI), germanium, and resist was described in detail, as well as the emitter definition with the trilevel resist system. The self aligned

PAGE 28

28 proces sing for emitter and base mesa wet etching, and base metallization definition was employed to reduce parasitic resistance and capacitance. Guard ring and ledge technique was used to preserve the current gain at small emitter dimensions and passivate the s idewall of the emitter mesa for eliminating the surface recombination. Moreover, dc characteristics and small signal rf performance were discussed for an effective 0.65 8.65 m 2 emitter area device. 2.2 Material Growth and Device Structure The paramete rs of the epitaxial layer structure for an InAlAs/InGaAsSb/InGaAs DHBT are listed in Table 2 1 The epilayers were grown on Fe doped semi insulating (100) InP substrates in a Riber 32P solid source molecular beam epitaxy (MBE) system equipped with arsenic phosphorus, and antimony valved cracker cells. Silicon and beryllium were the n type and p type dopants, respectively. The Sb 2 and As 4 flux w ere set to a beam equivalent pressure (B.E.P.) of 1. 4 10 7 torr and 8.0 10 6 torr to control the antimony a nd arsenic composition respectively The DHBT layer structure consists of a 300 nm n type In 0.53 Ga 0.47 As subcollector doped to 2 10 19 cm 3 a 150 nm n type In 0.53 Ga 0.47 As collector doped to 1 10 16 cm 3 a 44 nm p type In x Ga 1 x As 1 y Sb y based layer dope d with Be to 6 10 19 cm 3 a 40 nm n type In 0.52 Al 0.48 As emitter doped to 8 10 17 cm 3 a 5 nm n type InAlAs doped to 8 10 18 cm 3 a 30 nm n type In 0.53 Ga 0.47 As emitter cap doped to 2 10 19 cm 3 and a 4 nm n type InAs emitter cap doped to 8 10 19 cm 3 The substrate temperature was 490 o C throughout the entire growth except the base layer. The junction characteristics showed that the dopants were well confined. During the growth of the base layer, the substrate temperature was decreased to 450 o C to prevent beryllium out diffusion and antimony phase separation. 2.3 Small Scale InGaAsSb base Double Hetrojunction Bipolar Transistor Fabrication The schematic cross section of a triple mesa In 0.52 Al 0.48 As/In 0. 42 Ga 0. 58 As 0.7 7 Sb 0. 23 / In 0.53 Ga 0.47 As DHBT is illustrated in Figure 2 1. The layout of the unit cell DHBT was

PAGE 29

29 illustrated in Figure 2 2. The process flow s for InAlAs/InGaAsSb/InGaAs DHBTs fabrication followed the view directions of the lin a and line b in Figure 2 2 are illustrated schematically in Figure 2 3 and Figure 2 4 The first step is lift off definition of emitter contact metal. Processing of III V semiconductors is dominated by the use of the lift off technique rather than metal etching for the patterning of metal contacts [ 51 ]. This sit uation is largely due to the complicated multimetal contacts used for III V materials and the fact that many possible etchants used could attack the semiconductor. In the lift off technique, resist wa s spun onto the wafer, exposed, and developed to define the desired pattern. Metal deposition wa s then performed, and the resist subsequently removed, or "lifted off," to leave the metal present in the appropriate pattern. The original resist profile should have been slightly undercut to facilitate the metal lift off. In this process, there are two required conditions for patterning sub micron emitter contact fingers; one is the metal thickness and the other is the resolution for emitter patterns. In the final step of HBT processes, the benzocyclobutene (BCB ) was applied as the device passivation and planarization. The BCB etch back technique wa s needed to expose the emitter, base, and collector contact poles. In order to save a tolerance of emitter contact metal exposure during the later BCB etch back proc ess, a thick emitter metal stack around 6000 wa s necessary to be piled up. Meanwhile, a thick resist should be prepared for building a thick emitter metallization. By doing so, the feature resolution must be sacrificed and a sub micron emitter finger w a s very difficult to be achieved in the thick resist case due to the standing wave effect with optical lithography. In consideration of getting high resolution in sub micron scale and proper resist profile with perfect undercut for metal lift off, multila yer resist systems have been employed with electron beam (e beam) lithography to further reduce feature size (sub micron) and produce a correct resist profile for lift off of emitter metal. Prior to trilevel resist preparation, the e beam

PAGE 30

30 alignment marks for later e beam definition of emitter fingers were lift off defined on the fresh wafer surface with thin Ti/Au metal by using a contact mask. Trilayer resist system, including a thick PMGI bottom layer, silicon nitride or germanium transfer layer, and th e top light sensitive resist layer, has been widely used in lithography technique to tremendously improve the process for lift off definition of sub micron features with performing both high resolution and large resist thickness contrast [ 52 55 ]. However, it ha d drawbacks for using the transfer level of either silicon nitride or germanium film. For a low temperature deposited SiN x (<50C) transfer layer in the trilevel scheme, it allow ed for easy contact lithography and also allow ed for the use of an opti cal stepper due to its transparent property, but there wa s a severe charging effect for the e beam lithography to perform the misalignment, reduce the quality of feature shape and dimension, or even fail the e beam lithography operation. Ge film as a tran sfer layer in the trilevel scheme wa s thus an excellent candidate for e beam lithograph without the charging effect. However, the Ge transfer level wa s difficult to see through when using a contact printer, an optical stepper, or even an e beam writer. T his lead ed us to necessarily open the windows on this Ge film after Ge deposition for needs of seeing through and alignment. This step was discussed in detail in the following. To form the trilevel resist system, PMGI, based on polydimethylglutarimide, w as first spin coated on the surface of InAs emitter cap at 3000 rpm to achieve a thickness of 2.1 m, as a bottom layer of the tri layer scheme. The PMGI was then thermally cured on a hot plate at 200C for 20 min and it was reflowed to perform a planar c oating. Wafers were loaded into the electron beam evaporation system for the Ge deposition of 50 nm on the thermally stable PMGI beam al ignment marks were then defined with an optical lithography

PAGE 31

31 and developed by soaking in the Microposit MF 321 developer. Dry etching of the 50 nm thick Ge and 2.1 m thick PMGI bi layer was performed in a Technics Micro RIE (series 800 II) system. The Ge was dry etched in a CF 4 : O 2 = 9 : 1 discharge for 30 sec at a pressure of 180 mTorr (total gas flow rate 50 sccm), with 50 W of microwave power. The PMGI was etched in a pure O 2 discharge for 18 min with 50 W to expose the e beam alignment marks. The P MGI etching rate was 120 nm/min. Meanwhile, most of photoresist S1808 was simultaneously dry etched from the Ge surface in this O 2 reactive ion etching (RIE) process, and S1818 residue was able to be further removed from the Ge surface with acetone which did not dissolve the PMGI resist. For the e beam lithography of emitter fingers, the top layer in the trilevel resist system should be an electron beam resist. Polymethyl methacrylate (PMMA), the most commonly used e beam resist, was spin coated on the G e surface and baked at 180C for 20 min on a hot plate. Thus the tri1evel was completed and emitter finger definition was able to be performed. The process sequence of emitter contact metal definition with trilevel resists is illustrated schematically in Figure 2 5 The emitter fingers with sub micron feature size were defined on PMMA resist with Raith electron beam direct write system by the use of the e beam alignment marks. The small scaled rectangular emitter fingers of 0.6 9 2 0.8 9 2 1 9 2 1.2 9 2 and 1.4 9 2 were written by e beam. The developer is typically a solution of methyl isobutyl ketone (MIBK) and isopropanol (IPA) (1 : 3) to dissolved the e beam exposed area. Dry etching of the Ge and PMGI bi layers was just mentioned in the former paragraph, but a shorter O 2 RIE of only 16 min was employed to the PMGI etching for leaving a thin film of PMGI on the InAs surface. Since the surface of emitter cap could be damaged through the oxygen ion bombard after the PMGI overetch, a thin PMGI layer was intentionally left (shown in Figure 2 6 (a)) and removed with the MF 321 developer, as well as produced undercuts of >300

PAGE 32

32 nm in the PMGI, which was sufficient for the subsequent lift off step. After the remov al of thin PMGI layer with the MF 321 developer, the exposed semiconductor was generally still covered by residue which necessitated a post etch cleanup in an ultra violet ozone cleaning system (UVOCS Inc.). Figure 2 6 shows scanning electron microscope ( S EM) images of a feature etched into a PMGI layer using the Ge transfer layer as a mask. Immediately after the MF 321 developing, there were still some remnant products on the semiconductor surface (shown in Figure 2 6 (b)), but these residue were removed from the InAs surface after an additional ozone treatment, as shown in Figure 2 6 (c)). The Ge transfer layer thus fulfilled our requirements to have good adhesion on the underlying PMGI and the ability to selectively be removed and expose the PMGI for p atterning. After e beam definition, the emitter contact was patterned with Ti/Au (200/6000 ) metallization and lifted off by soaking the wafer in the MF 321 developer, which was commonly used to dissolve the PMGI. An emitter specific contact resistivity of 4.6 10 7 ohm c m 2 was obtained from transmission line measurements (TLMs) performed on 100 100 m 2 pads separated by 2, 4, 8, or 10 m. SEM micrographs of 600 nm as deposited Ti/Au were shown in Figure 2 7 (a) and (b), and the lift off process was easily accomplished using MF 321 developer, as shown in Figure 2 7 (c). Emitter metal was then used as a mask for self aligned wet etching to expose the base layer. With the emitter metal contact in place, wet chemical etching of the emitter mesa was per formed with selective wet etchant Citric acid/H 2 O 2 for InAs and InGaAs, followed by wet etchant H 3 PO 4 /H 2 O 2 /H 2 O for InAlGaAs and InAlAs and selective wet etchant HCl : H 2 O for InAlAs. The etch selectivity is one of the most critical requirements for HBT fa brication. The selectivity for InAs or InGaAs over a graded InAlGaAs was 25 : 1 and an etch rate of In(Ga)As

PAGE 33

33 in the acid solution of Citric acid : H 2 O 2 = 2 : 1 was 3.5 nm/sec. Wet chemical etching for InAlGaAs and InAlAs over InGaAsSb was performed in th e mixtures of H 3 PO 4 : H 2 O 2 : H 2 O = 1 : 1 : 100 with an etch rate of 1.5 nm/sec. H 3 PO 4 /H 2 O 2 /H 2 O solution only showed a low selectivity of <5 for InAlAs/InGaAs(Sb) material system, however, the etch rate was slow and could be easily handled and controlled i n the process. A thin InAlAs layer of 200 was preserved for performing the ledge on the base layer. Therefore, H 3 PO 4 /H 2 O 2 /H 2 O wet etching needed to be controlled carefully and stopped on the 200 InAlAs emitter layer over InGaAsSb base. This wet etch process could be easily achieved by recognizing the color of the wafer surface to turn into red while all of InAlGaAs was removed and red InAlAs emitter layer was exposed. The InAlAs emitter was then wet etched continuously for ~15 sec to leave around 20 0 over InGaAsSb base layer. During the wet etching, more than 90 nm undercuts of the emitter mesa below the emitter metal were obtained. The top view SEM image of a 0.6 9 2 emitter finger was shown in Figure 2 8 (a), and the undercuts was observed during emitter mesa wet etching, as shown in Figure 2 8 (b). A smallest width of 250 nm for the emitter mesa was achieved using a 0.6 9 2 emitter metal as a mask. The effe ctive widths of 0.25, 0.45, 0.65, 0.85, and 1.05 m were obtained for 0.6 9 2 0.8 9 2 1 9 2 1.2 9 2 and 1.4 9 2 emitter fingers, respectively. After the ledge and guard rings processes, the wet chemical selective etch in mixtures o f HCl : H 2 O = 4 : 1 was used for etching the rest of InAlAs over InGaAsSb. The excellent etch selectivity for InAlAs over InGaAsSb was over 2000 and an etch rate was 10 nm/sec. Guard rings were formed around the emitter periphery in order to preserve the current gain at small emitter dimensions. The process sequence of the front side device for guard ring and ledge fabrication is illustrated schematically in Figure 2 9 (top). Wafers with an InAlAs

PAGE 34

34 emitter layer of 200 were loaded into the Plasma Therm plasma enhanced chemical vapor deposition (PECVD) system for the deposition of silicon nitride. The temperature of the system 100 standard cubic centimeters pe r minute (sccm), 27 sccm, and 400 sccm, respectively. The pressure in the chamber was maintained at 900 mTorr with a forward rf power of 30 W. Sidewall passivation of emitter mesa and emitter metal stacks by PECVD SiN x of 200 was then performed. The S iN x provided conformable coverage over the emitter contact, and then CF 4 RIE dry etching was used for an etch back step to expose the thin InAlAs emitter layer and the emitter metal. The SiN x on the area vertically projected below the emitter metal, as a mask, was not dry etched by CF 4 RIE. SiN x guard rings protected the sidewall of emitter mesa and the area covered with SiN x under the emitter metal from the wet chemical selective etch in HCl/ H 2 O solution, and thus a ledge was performed and InGaAsSb base was exposed, as shown in Figure 2 9 (bottom). Prior to the lift off definition of base contact metal with the optical lithography, bi layer resists of MicroChem LOR 3A and Microposit S1808 were spin coated on the wafer at 4000 rpm for 30 sec for both resi respectively. Resist LOR has been widely used to perform an excellent undercut layer in bi layer lift off processing. The base metal was deposited using an e beam evaporator. Pt/Ti/Pt/ Au base metallization was used with a total thickness of 900 (less than the emitter mesa thickness of 1200 ) to prevent the shorting due to the metal connection between emitter and base metal stacks. Base metal was then self aligned around emitter mesa with the emitter metal as a mask. Soaking the wafer in acetone was used to lift off the base metal, leaving an unshorted, well defined base contact. This simple self aligned process enabled narrow separation between the

PAGE 35

35 emitter contact mesa and base ele ctrodes. The first Pt layer in the base metal stacks can make a better Ohmic contact to p type InGaAsSb layer with its higher work function of 5.65 eV, compared with Ti work function of 4.33 eV. Pt/Ti/Pt/Au metallization was thermally treated at r 1 min in a tube furnace with flowing nitrogen introduce to provide a specific contact resistivity of 3.7 10 7 ohm c m 2 to this p + base layer. Self aligned base mesa etching was performed using a wet etchant H 3 PO 4 : H 2 O 2 : H 2 O = 1 : 1 : 20 with an etch rate of 3 nm/sec to achieve a 194 nm base mesa and reach the InGaAs sub collector layer. Microposit S1818 photoresist was patterned with the contact mask to cover the emitter active area and the base metal was provided as a mask during base mesa wet etch process. Ti/Au metallization employed for the collector contact (5 6 10 6 ohm c m 2 contact resistivity) with lift off definition. The sheet resistance and specific contact resistivity for emitter, base, and collector, measured using transmission line me asurement (TLM), were listed in Table 2 2. Device isolation was proceeded to further reduce the current leakage with isolating the active regions to another and increase the breakdown voltage of HBTs; it was achieved using mesa wet etching with the select ive wet etchant H 3 PO 4 /H 2 O 2 /H 2 O for InGaAs over InP substrate. The etch rate to InP was so slow that this wet etch process could stop on the InP substrate. 1 m Air bridge, which connected the base contact in the active area and the contact pole of base i nterconnected metal pad, was formed with the base metal deposition. InGaAsSb base and InGaAs collector/sub collector layers below this 1 m Air bridge were removed with the wet chemical etching of base mesa and device isolation. T he current gain cutoff f requency ( f t ) and power gain cutoff frequency ( f max ) equations are as follows,

PAGE 36

36 a nd where base collector C jc C j e R base R emi R coll I c k, T, q are the base transit time, the collector transit time, the capacitan ce between base and collector, the capacitance between base and emitter, the base resistance, the emitter resistance, the collector resistance, the collector current, the Boltzmann constant, the temperature, and the electron charge, respectively. T he capa citance between base and collector ( C j c ) plays a very importance role in the high frequency performance. The C j c can be reduced and cut off frequency can be increased with an air bridge design due to removal of materials under the air bridge. During base mesa wet etching, undercuts of ~80 nm on both sidewalls below the air bridge was performed with a side etch rate of 1.2 nm/sec. Another 120 nm undercut on each sidewall was produced while InGaAs in the isolation regions was completely removed during the device isolation wet etch process. There were remnant materials of ~600 nm in width under this 1 m wide air bridge, and thus the wafer was kept soaking in the acid solution to empty all materials under the bridge for additional 4 ~ 5 minutes. Figure 2 1 0 (a) and (b) showed the air bridge without any materials below before and after removal of the photoresist which was used as a protection on the active regions, respectively. The SEM micrograph of one unit device after stripping the resist was shown in F igure 2 11 Benzocylcobutene (BCB, DOW Cyclotene) of 2.5 m thick, serving as planarization and insulating dielectric materials in silicon and III V compound semiconductor microelectronic fabrication, was spin coated at 4000 rpm for 30 sec, after the adhe sion promoter (DOW AP3000) was formerly spin coated for improving the adhesion of BCB to other different substrates. BCB coating was then thermally cured in a tube furnace continuously introduced with nitrogen gas.

PAGE 37

37 The cure temperature was increased from solid the BCB coating. After planarizationthe, in order to expose the emitter metal, as well as base and col lector contact poles, BCB etch back process was accomplished by means of plasma etch processes. A mixture of CF 4 and O 2 (CF 4 /O 2 ratio = 9) was used in Technics Micro RIE (series 800 II) system to perform a controlled etch rate of 0.1 m/sec at a pressure of 180 mTorr (total gas flow rate 50 sccm), with 50 W of microwave power. The SEM micrographs of a single emitter finger exposing ~300 nm emitter metal over the BCB plane, as well as the exposed base and collector contact poles were shown in Figure 2 12 ( a) and (b), respectively. Ti/Au interconnect metal for emitter, base, and collector electrodes was then defined by lift off, as shown in Figure 2 13 2. 4 Small S cale InGaAsSb base Double Hetrojunction Bipolar Transistor Electrical Characterization The de vice dc parameters were measured with an Agilent 4156B parameter analyzer. The common emitter breakdown voltages (BV CEO ) at an open base were 2.57 V at J C = 1 kA/cm 2 and 2.63 V at J C = 10 kA/cm 2 with a collector doping of 1 10 16 c m 3 and a thickness of 150 nm for an effective 0.65 8.65 m 2 emitter area device. Common emitter current voltage ( I V ) characteristics were shown in Figure 2 14 Typical room temperature Gummel characteristics and dc current gain of the In 0. 42 Ga 0. 58 As 0.7 7 Sb 0. 23 DHBT were sho wn in Figure 2 15 (top). The forward turn on voltage of the In 0. 42 Ga 0. 58 As 0.7 7 Sb 0. 23 DHBT decreased to as low as 0.38 V at the current density of 1 /cm 2 as compared to the turn on voltage of 0.505 V and ~0.45 V for the conventional InP/InGaAs SHBT and D HBTs at the same current density [ 15 1 8 1 9 ]. The collector and base current ideality factors were C = 1.28 and B = 1.41, respectively. The low value of B confirmed the improvement of the replacement of the type II E/B junction by the

PAGE 38

38 type I E/B junc tion, which helped to alleviate the electron pileup and reduces the tunneling recombination current. The maximum dc current gain was 123.8 at the current density of 364.73 kA/cm 2 ( V CB = 0 V), as shown in Figure 2 15 ( bottom ), plotted with log scaled dc ga in versus collector current. The on wafer s parameters of the devices between 50 MHz and 40 GHz were measured using an HP8722C vector network analyzer. Both the current gain frequency f T and maximum oscillation frequency f max were determined by the 20 dB/decade extrapolation from the common emitter current gain h 21 U curves, respectively. The frequency dependence of the current gain and power gain, which followed the 20 dB/decade extrapolation very well, was shown in Figure 2 16 The DHBT having an emitter area of 0.65 8.65 m 2 showed f T / f max = 260/485 GHz at J C = 302 kA/cm 2 and f T / f max = 268/473 GHz at J C = 508 kA/cm 2 The maximum of f max was exhibited around 485 GHz, which was governed by the collector c apacitance and base resistance. InGaAsSb base with a high doping concentration of 6 10 19 cm 3 low specific contact resistivity of 3.7 10 7 cm 2 The time constant R B (C jC,i + C jC,x ) was tremendously reduced with a more efficient InGaAsSb base with low resistance and t he self aligned processing to minimize parasitic resistances and capacitances. Guard rings and the ledge designed on the InGaAsSb DHBT helped to suppress generation of the surface recombination current and also dramatically reduce the resistance from the base contact to emitter active regions to improve the frequency of power gain. The dependence of f T and f max on collector current density at 1 V collector bias was shown in Figure 2 17 Also, based on the figure of merit (F.O.M.) for D type Flip Flop ICs defined as f T f max / ( f T + f max ) [ 20 ], Figure 2 18 compared the

PAGE 39

39 f T and f max of our InGaAsSb DHBTs with those taken from literature [ 13 31 ]. UF DHBT with 0.65 8.65 m 2 emitter area has delivered a comparable performance among InP base HBTs. Table 2 1 Epitaxial layer structure parameters of fabricated InAlAs/InGaAsSb/InGaAs DHBTs. Layer Material Carrier concentration ( cm 3 ) Thickness (nm) Dopant Emitter Cap n + InAs 810 19 4 Si E.C. graded n + In 0.53 Ga 0.47 2 ~ 810 19 16 Si Emitter cap n + In 0.53 Ga 0.47 As 210 19 30 Si E.C. graded n + InAlGaAs 810 18 25 Si Emitter cap n + In 0.52 Al 0.48 As 810 18 5 Si Emitter n In 0.52 Al 0.48 As 810 17 40 Si Base p In 0.42 Ga 0.58 As 0.7 7 Sb 0. 23 610 19 44 Be Collector n In 0.53 Ga 0.47 As 110 16 150 Si Subcollector n + In 0.53 Ga 0.47 As 210 19 300 Si Semi insulating InP substrate Table 2 2 Sheet, contact, and transfer resistances for In 0.52 Al 0.48 As/In 0.42 Ga 0.58 A s 0.7 7 Sb 0. 23 / In 0.53 Ga 0.47 As DHBT structure. Transfer Resistance R t ( mm) Sheet Resistance R sh ( Specific Contact Resistivity C ( cm 2 ) Transfer Length L T (m) Emitter 0.033 24.1 4.6010 7 2.76 Base 0.211 1206 3.7010 7 0.35 Collector 0.049 4.2 5.6010 6 23.09

PAGE 40

40 Figure 2 1. Schematic cross sectional view of a fabricated double mesa In 0.52 Al 0.48 As/ In 0.42 Ga 0.58 As 0.7 7 Sb 0. 23 /In 0.53 Ga 0.47 As DHBT.

PAGE 41

41 Figure 2 2. The device layout of a unit cell. Base air bridge Emitter metal Base metal Collector metal Isolation Contact window for base Contact window for collector Emitter Base Collector Isolation a b

PAGE 42

42 Figure 2 3 Schematic of process sequence for In 0.52 Al 0.48 As/In 0.42 Ga 0.58 As 0.7 7 Sb 0. 23 / In 0.53 Ga 0.47 As DHBT followed the view direction of the line a in Figure 2 2).

PAGE 43

43 Figure 2 4 Schematic of process sequence for In0. 52Al0.48As/In0.42Ga0.58As0.7 7 Sb0. 23 / In0.53Ga0.47As DHBT from the front side view (followed the line b in Figure 2 2)

PAGE 44

44 Figure 2 5 Schematic of process sequence for emitter contact metal definition on trilayer resist system. O 2 plasma etching off PMGI PMGI InAs Ge PMMA Developer solution removes the residue Get the undercuts of PMGI PMGI InAs Ge Define emitter fingers by E beam CF 4 p lasma etching off Ge PMGI InAs PMMA Ge Deposit emitter metal PMGI InAs Ge Ti/Au

PAGE 45

45 Figure 2 6 SEM micrographs of trilevel resist structure after CF 4 /O 2 RIE (a), after soaking in developer (b), and after UV ozone treament (c). PMGI thin film Ge 2 m (a) 300 nm PMGI Ge (b) 300 nm PMGI Ge (c)

PAGE 46

46 Figure 2 7 SEM micrographs of as deposited emitter metal (a and b) a nd emitter metal after lift off process (c). 300 nm (c) 300 nm 600nm emitter metal deposition (b) 2 m (a)

PAGE 47

47 Figure 2 8 The top 2 emitter finger (a) and the front side of emitter fingers with different widths of 0.6, 1, 1.2 m (b). 300 nm 0.6 m 300 nm 1 m 300 nm 1.2 m (b) 1 m (a) 0.6 x 9 m 2

PAGE 48

48 Figure 2 9 The process sequence of guard rings and ledge fabrication (top) and the SEM micrograph of fabricated ledge structure (bottom). Guard ring and Ledge Emitter metal Base layer 100 nm

PAGE 49

49 Figure 2 10 SEM micrographs of 1 m air bridge with photoresist after isolation wet e tc hing (a) and after the resist strip (b). Air B ridge Photoresist Isolation mesa 2 m (a) 1 m (b) Emitter

PAGE 50

50 Figure 2 11 SEM micrograph of 0.65 8.65 m 2 device prior to BCB planarization. Figure 2 1 2 SEM micrographs of exposed emitter finger (a) and collector/base contact poles (b) after BCB etch back pro cess. 10 m C E B 10 m Collector contact pole Base contact pole Emitter finger (b) 1 m Emitter finger (a)

PAGE 51

51 Figure 2 1 3 The top view of optical microscope image for a fabricated device after Ti/Au interconnect metal deposition. Figure 2 1 4 Common emitter characteristics of the In 0.52 Al 0.48 As/In 0.42 Ga 0.58 As 0.7 7 Sb 0. 23 / In 0.53 G a 0.47 As DHBT. Collector Base Emitter Emitter

PAGE 52

52 Figure 2 1 5 Gumm 0.52 Al 0.48 As/In 0.42 Ga 0.58 As 0.7 7 Sb 0. 23 / In 0.53 Ga 0.47 current (bottom).

PAGE 53

53 Figure 2 1 6 Current gian h 21 gain U rf extrapolations of 0.65 8.65 m 2 DHBT.

PAGE 54

54 Figure 2 1 7 Dependence of f T and f max on collector current density J C for the In 0.52 Al 0.48 As/ In 0.42 Ga 0.58 As 0.7 7 Sb 0. 23 /In 0.53 Ga 0.47 As DHBT with a 0.65 8.65 m 2 emitter area. F igure 2 1 8 Overview of f T and f max data for high speed InP and SiGe based HBTs. The figure of merit (F.O.M) was defined as f T f max / ( f T + f max ).

PAGE 55

55 CHAPTER 3 EFFECTS OF OHMIC METAL ANNEALING ON INALN/GAN HIGH ELECTRON MOBILITY TRANSISTOR STRUCTURES 3 .1 Background InAlN/GaN is an emerging materials system with high thermal and chemical stability and lattice matching to GaN at an In mole fraction of 0.17 [ 56 67 ] The absence of strain in the barrier layer suggests that InAlN/GaN High Elect ron Mobility Transistors (HEMTs) might have superior device reliability as compared to more conventional AlGaN/GaN HEMTs [68] as some reports suggest that reliability issues in the AlGaN/GaN devices are related to lattice defects caused by the inherent st rain due to the lattice mismatch in this system [69] Promising dc, rf and output power performance has been reported for InAlN/GaN HEMTs on Si, sapphire and SiC substrates [56,57,59 65] However, there are still issues to be explored because the growth temperature of InAlN is lower than the growth temperature of GaN and one might expect consequences for the conditions under which Ohmic contacts are formed on the InAlN/GaN heterostructure. An example is the annealing temperature for the commonly used Ti/ Al/Ni/Au Ohmic metallization on InAlN/GaN HEMTs. The reported annealing temperatures for Ohmic contacts on these structures have varied from t C [60] to a single anneal at either 800 [ 67 ] [ 66 ] In this paper we pro vide a systematic investigation of the effect of Ohmic contact annealing temperature on the stability of electrical properties, surface and interface morphology and subsequent device performance of InAlN/GaN HEMTs. The optimum annealing temperature to ach ieve minimum contact resistance is 800 conditions there is no significant drop in electron mobility in the heterostructure and only a small reduction (~10%) in sheet carrier density. However, clear evidence of interfacial roughening and Ga out diffusion accompanies annealing at higher temperatures.

PAGE 56

56 3 .2 Experimental The HEMT structures were grown with by Metal Organic Chemical Vapor Deposition (MOCVD), starting with a thin AlGaN nucleation layer, 1.9 m carbon doped GaN buffer layer, 50 nm undoped GaN layer, 10.2 nm undoped I nAlN layer and finally, in some cases, a 2.5 nm undoped GaN cap 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 10 13 and 2.0 10 13 cm 2 respectively, with and without the GaN cap. The corresponding electron mobilities were 1000 and 1350 cm 2 / V s. A schematic of the structures both with and without the GaN cap is shown in Figure 3 ray scans confirmed the close lattice matchi ng of the InAlN and GaN layers in both structures. To examine the effect of annealing temperature on the electrical properties of the samples, the as grown samples were annealed from 700 density obtained from Hall measurements. The morphology changes as a result of the annealing were measured by Atomic Force Microscopy (AFM), while changes in interfacial roughness between the InAlN and GaN were investigated with x ray reflection (XRR) and diffuse x ray scattering. The thickness, mass density and surface and interface roughness were estimated by simulating with the aid of a commercial software package (WinGixa from Panalytical) the x ray a 2theta rocking curves were also acquired with the same instrument but using Cu K radiation (four bounce hybrid monochromator on the primary optics and a 1 mm slit in front of the detector on the secondary optics). The rocking curves were simulated with the aid of the Epitaxy software from Panalytical to extract information about the elemental composition of the In x Al 1 x N layer and its thickness.

PAGE 57

57 The contact properties were obtained from transmission line measurements (TLM) performed on 100 m 2 pad R C was obtained from the relation [ 70 ] R C = ( R T S d / Z )/2, where R T is the total resistance between two pads, S is the sheet resistivity of the semiconductor under the contact, d is the pad spacing, and Z is the C is then obtained C = R C L T Z where L T is the transfer length obtained from the intercept of a plot of R T vs d The Ohmic contacts were formed by e beam evaporation of TiAlNiAu patterned by liftoff. These contacts were also annealed at temperatures from 700 850 C for 30 sec under flowing N 2 ambient. Finally, HEMTs were also fabricated, using liftoff of Ni/Au gate contacts and Cl 2 /Ar Inductively Coupled Plasma etching for mesa isolation. The dc characteristics of these devices were measured on a parameter analyzer. 3 .3 Annealing Temperature Dependence of Ohmic Contact Resistance and Morphology on InAlN/GaN High Electron Mobility Transistor Structures Figure 3 2 shows the change in both Ha ll mobility (a) and sheet carrier density (b) as a function of annealing temperature for heterostructures with and without the GaN cap. No degradation of electron mobility was observed for the samples annealed at 700 and 750C, but there was a significant degradation (35%) of mobility for the sample annealed at 850 C. Note that the decrease in mobility is accompanied by a decrease in sheet carrier density, indicating that electrical compensation or material degradation is the cause of the change in properti es. A summary of the Ti/Al/Ni/Au Ohmic contact properties for these same anneal temperatures are shown in Figure 3 3. An optim al annealing temperature at 800 C was obtained using the TLM results. The minimum transfer resistance for the contacts of 0.65 o hm mm (specific contact resistivity of 2 10 5 o hm cm 2 ) was achieved after 800 structures without the GaN cap. The presence of the GaN always leads to a slightly higher

PAGE 58

58 contact resistance and pushed the optimum anneal temperature to higher values, as expected since that is more thermally stable than the InAlN and wou ld react with the metal at a higher temperature. There was no significant dependence of the contact resistance on measurement ort mechanism is field emission [ 71 72 ]. Thermionic emission would have significant temperature dependence and thermionic field emission is operative at lower doping ranges (10 16 10 18 cm 3 ). Figure 3 4 shows Ohmic metal topography after annealing at different temperatures for the InAlN/GaN HE MTs without the 2.5 nm thick GaN cap. The surface shows a reacted appearance beginning at 750 temperature. The same basic trends were observed for samples with the GaN cap, as shown in Figure 3 5, with the onset of roughening shifted to higher temperatures. In this contact scheme, the Ti/Al bi layer is the key for forming an Ohmic contact through formation of a TiN x phase while the Ni/Au over layer is mainly for preventing out diffusion and reducing morphology degradation. The Au also reduces the sheet resistance of the layers and prevents oxida tion during high temperature annealing. To understand more about the effect of annealing on the structural properties of the heterostructure, samples were examined by XRR and diffuse x ray scattering before metallization. Figure 3 6 (a) displays the x ra y reflectivity curves acquired from a control interference fringes were dampened while their period shortened. Simulations of these XRR curves showed that the interfa ce width between the substrate and the InAlN layer increased from 0.4 nm to 1.0 nm, while the interface width between the InAlN and the GaN capping layer also increased from 0.5 to 0.8 nm. Since XRR can not distinguish between a morphological rougher inter face and a graded one caused by inter diffusion between layers, we also performed diffuse

PAGE 59

59 scattering measurements. The results clearly showed that the morphological roughness of the interfaces increased after annealing. Table 3 1 summarizes the thicknesse s of the InAlN and GaN cap layers before and after annealing. It was also surprising to measure that the density of the InAlN layer increased from a value of 3.90 g/cm 3 for the control sample (very close to the density of 3.85 g/cm 3 that one obtains for x = 3 after annealing. Nevertheless, the RC simulation gave the same x = 0.16 value for the InAlN layer. This suggests that Ga diffusion inside the InAlN layer is the explanation for the higher mass dens ity after annealing, as well as its increased thickness. The decrease of the thickness of the GaN capping layer also supports the out diffusion of Ga atoms as the main change during annealing. Figure 3 6 (b) displays the omega 2theta rocking curves acquire d around the (002) diffraction line. Note the similar decrease in the amplitude for thickness fringes after the annealing, concomitant with a slight increase of their period, very similar to the XRR results. The red trace is a simulation curve that was ge nerated to obtain the In fraction (x = 0.16) and the thickness of the layers. The position of the diffraction line corresponding to the InAlN layer did not change after anneal, implying a similar x value. To examine the role of contact annealing condition s on the resulting HEMT performance, devices were tested after annealing at different temperatures. Figure 3 7 shows the saturation drain current characteristics for zero gate voltage on the InAlN/GaN HEMTs both without cap and with the 2.5 nm GaN cap. No sec annealing where the contact resistance is lowest. The lower contact resistance is also evident in the higher slope of the I V characteristics in the quasi linear range.

PAGE 60

60 Figure 3 8 (a) shows the drain I V characteristics from +4V to 4V gate bias for devices with and without the GaN cap. In both cases, the contacts were formed by annealing at the The devices without the cap show slightly higher drai n currents, with a maximum near 1.3 A/mm, certainly very competitive with more conventional AlGaN/GaN HEMTs. The maximum transconductance was 366 mS/mm, as shown in Figure 3 8 (b). The one place where the GaN cap proved to be advantageous was in increasi ng the effective barrier height for gate metals. Figure 2 9 shows the gate I V characteristics for Ni/Au Schottky contacts on the HEMTS with and without the GaN cap. Results are shown for two different devices with the cap to give an idea of the uniformi ty of the data. There is a clear reduction in forward current at a given bias in the case of the capped structures, due to a higher barrier height of 1.01 eV compared to 0.91 eV on the InAlN layer. Table 3 1. Thickness of the layers estimated from the s imulation of the XRR and RC measurements. Sample InAlN cap thickness (nm) XRR InAlN cap thickness (nm) RC GaN cap thickness (nm) XRR GaN Cap thickness (nm) RC Control 10.7 11.0 1.2 1.5 Annealed 11.9 11.9 1.1 0.9 Figure 3 1 Schematic of the two InAlN/GaN layer structures investigated, ie. with (a) and without (b) a thin GaN cap. (a) ( b )

PAGE 61

61 Figure 3 2 Variation of mobility (a) and sheet carrier density (b) as a function of annealing temperature for structures with and without a Ga N cap layer.

PAGE 62

62 Figure 3 3 Contact resistivity (a), sheet resistance (b) and transfer resistance (c) for contacts on both types of InAlN/GaN structures as a function of annealing temperature.

PAGE 63

63 Figure 3 4 Optical m icrographs of contacts on InAlN/GaN structures without the GaN cap, as a function of annealing temperature. o o o o

PAGE 64

64 Figure 3 5 Optical micrographs of contacts on InAlN/GaN structures with the GaN cap, as a function of annealing temperatur e. o o o o

PAGE 65

65 Figure 3 6 (a) XRR curves acquired from an as grown sample and from a sample that was annealed at 850 o C ; (b) High resolution omega 2theta rocking curves acquired near the (002) GaN diffraction line for control (blue trace) and annealed sample (green trace); the red trace is the simulated curve for the control sample. (a) ( b )

PAGE 66

66 Figure 3 7 Drain I V characteristics at zero gate voltage for InAlN /GaN HEMTs without (a) or with (b) GaN cap layers, as a function of contact annealing temperature.

PAGE 67

67 Figur e 3 8 (a) Comparison of drain I V characteristics at 4 V gate voltage for InAlN /GaN HEMTs with or without GaN cap layers, for the optimized contact annealing temperature of 800 function of gate voltage.

PAGE 68

68 Figure 3 9 Log plot of forward and reverse I V characteristics from HEMTs with and without GaN caps. The barrier heights were extracted from the forward I V characteristics. Results from two devices with GaN caps are shown to give an idea of the variation seen from device to device.

PAGE 69

69 CHAPTER 4 PASSIVATION ON GAN BASED ELECTRONIC DEVICES 4 .1 Passivation of AlN/GaN high electron mobility transistor using o zone treatment 4.1.1 Background Wide band gap AlGaN/GaN high electron mobility transistors (HEMTs) have been recognized as excellent candidates for RF/microwave power amplifiers because of their high power and high speed handling capabilities [ 73 82 ]. Ho wever, the large band gap of AlN (6.2 eV) provides even better carrier confinement and lowers the gate leakage current, and the absence of alloy disorder (compared to AlGaN barriers) results in improvement of both low and high field carrier transport. The output current of the HEMT is normally limited by the carrier sheet density and carrier injection velocity; therefore, a high carrier density, along with high carrier mobility, is desired. A number of variations on AlN/GaN HEMTs have been explored with no table success. However, AlN can be easily etched by the standard photoresist developer solution employed during the device fabrication. A protective layer deposited on the surface is needed to protect the AlN and act as a gate insulator. Silicon nitride (SiN x ) deposited by plasma enhanced vapor deposition or electron beam deposition has traditionally been used to protect the AlN [ 83 87 ] Ozone treatment prior to the standard SiN x process has been shown to be an efficient and effective method to improv e AlGaN/GaN HEMT device isolation characteristics [ 88 90 ] Recently, we have employed an ozone treatment to oxidize the AlN surface of the AlN/GaN HEMTs; this oxidized AlN surface then protected the underlying AlN during the device fabrication. An addit ional O 2 plasma treatment was used to further oxidize the AlN surface layer at the gate area of the AlN/GaN HEMT to control the threshold voltage of the device [ 91 ]

PAGE 70

70 In this work, we report on the effect of ozone and oxygen plasma treatments on AlN/GaN HE MT gate diode and drain current voltage ( I V ) characteristics as well as drain current recovery during the gate lag measurements. Energy Dispersive X ray Spectroscopy (EDS) was also used to analyze the composition of the AlN surface on the AlN/GaN HEMT sa mple before and after the ozone and oxygen plasma treatment. 4 .1 2 Experimental AlN/GaN HEMT structures were grown on c plane sapphire in a M olecular B eam E pitaxy ( MBE) system equipped with an RF nitrogen plasma source from SVT Associates (SVTA). During growth, reflection high energy electron diffraction was used to monitor surface morphology. Other in situ measurements, including emissivity corrected surface temperature, thin film growth rate, and III/V flux ratio were performed by a combination of pyrom etry and two color reflectometry (SVTA IS4000). The growth process started with surface nitridation at high temperatures using the rf nitrogen plasma source, followed by the growth of a thin AlN nucleation layer. Next, a 2 to 3 m low defect GaN buffer was grown A relatively low TD density (< 1 10 8 cm 2 ) in these films has been previously confirmed by both etch pit density measurements, using atomic force microscopy (AFM), and by transmission electron microscopy (TEM) [ 90 ]. Finally, the AlN/GaN active layer was formed by growing a thin (<5 nm) AlN layer at about 700 C. The optical and electrical properties were characterized by cathode luminescence, Hall, and capacitance voltage measurements. The mobility and sheet carrier density in the two dimensional electron gas channel ranged from 1300 1900 cm 2 /V s and 1.5 3.5 10 13 cm 2 respectively. The typical sheet resistance of the sample was below 200 / Prior to the normal device fabrication, the samples were treated w ith a n UV generated o zone for 1 min to protect the AlN surface layer. The ozone treated surface exhibited resistance

PAGE 71

71 to positive resist developer solution, which is known to easily attack the AlN. Device fabrication began with mesa isolation by Cl 2 /Ar i nductively coupled plasma (ICP) etching (150 W source power, 40 W RF chuck power). Isolation currents were less than 2 A at 40 V bias for a mesa depth of 1 2 00. Ohmic contacts were formed by lift off of e beam deposited Ti/Al/ Ni /Au based metallization, and subsequently annealed at 8 50C for 30 s under a N 2 ambient. The gate fingers with a dimension of 40 0 nm 200 m were defined using a Raith electron beam direct write system. Prior to the gate metal deposition, oxygen plasma was used to further oxidi ze the AlN into Al oxides to adjust the threshold voltage. The oxygen plasma treatment was performed using a parallel plate system with 18 W rf power for 20 sec. E beam deposited Ti/Au was employed for the gate metallization The distance between the sour ce and drain was 2 m. A schematic cross section view of the device is shown in Figure 4 1. The DC characteristics of the HEMTs were measured with a Tektronix curve tracer 370A and an HP 41 56 parameter analyzer The HEMT dc parameters were measured in dc and pulse mode at 25 C, using a parameter analyzer for the dc measurements, and a pulse generator, a dc power supply and an oscilloscope for the pulsed measurements. For the gate lag measurements, the gate voltage, V G was pulsed from 5 V to different vo ltages at different frequencies with a 10% duty cycle. 4 .1 3 Results and Discussions Figure 4 2 shows the EDS scans of the AlN/GaN HEMT surface before and after the ozone treatment as well as the EDS scan of the surface after the additional oxygen plasm a exposure. The top EDS spectrum is the full spectrum of the AlN/GaN HEMT surface treated with ozone and the additional oxygen plasma. The oxygen peak was minimal and could only be seen in the enlarged EDS scan. Thus, the ozone generated oxide must exis t as a few monolayers

PAGE 72

72 on the AlN surface since the electron beam voltage of 3 k eV used in EDS scans had an interaction depth around 100 nm. The dominant peaks in the EDS spectrum were Al, Ga, and nitrogen. The enlarged EDS spectrum of the AlN before the ozone treatment is also illustrated in Figure 4 2; there was no oxygen peak observed. The intensity of the oxygen peak for the AlN/GaN HEMT treated with both ozone and additional oxygen plasma was slightly higher than that of ozone treated sample. Figure 4 3 shows the forward and reverse gate current of the AlN/GaN HEMTs treated with ozone of the d mode HEMT and an additional oxygen plasma exposure of the e mode HEMT. There was no forward Schottky turn on observed and a metal oxide semiconductor (MOS) di ode like I V was obtained. This is consistent with the EDS results that ozone generated oxide served as the gate oxide. The gate current of the e mode HEMT was one order of magnitude lower than that of the d mode HEMT. This implies that a thicker Al oxi de layer formed in the gate area with the additional oxygen plasma treatment. The drain current characteristics and dc transfer characteristics for both e mode and d mode AlN/GaN HEMTs are illustrated in Figure 4 4 and Figure 4 5, respectively. T he e mod e and d mode HEMTs exhibited sharp pinch off voltages at 0.5 V and 3.5 V respectively The threshold voltage shifted from 2.76 V to 0.13 V to turn the d mode device to e mode device after additional 18 sec oxygen plasma treatment. The ozone and oxygen plasma generated oxide allowed the gate to be operated at high voltage without significant gate leakage current. The maximum saturation drain currents of the e mode and d mode HEMT were 0.5 and 1.25 A/mm at gate bias voltages of 6 V and 4 V, respectively The gate lag data of the drain current as a function of pulse gate voltage for the e mode and d mode HEMTs are shown in Figure 4 6 The gate lag measurements were conducted at a

PAGE 73

73 fixed drain voltage of 10 V, and the gate voltage was pulsed from 5 V t o different gate voltages at a frequency of 1 kHz, 10 kHz, or 100 kHz with a 10% duty cycle. For the d mode HEMT, there was no dispersion of the drain current observed regardless of the frequency of the gate pulses. This indicates that the ozone passivati on did not introduce defects on the AlN/GaN HEMT surface. For the e mode HEMT, there was no drain current reduction for the gate lag measurements at 1 kHz and 10 kHz. However, the e mode HEMT showed 5% drain current degradation during the pulse measureme nt performed at 100 kHz; this implied that there was some damage introduced under the gate metal during the oxygen plasma treatment. 4 2 Effect of Silicon Nitride Passivation on Isolation Blocking Voltage in AlGaN/GaN High Electron Mobility Transistor St ructure 4.2.1 Background AlGaN/GaN High Electron Mobility Transistors (HEMTs) show great promise for applications ranging from high frequency wireless base stations and broad band links to commercial and military radar and satellite communications [ 92 94 ] The development of proper device passivation for the AlGaN/GaNHEMTS is critical to device stability and reliability [ 95 102 ]. The effect of the surface traps on the drain current collapse during large signal operation can be alleviated by the use of ap propriate passivation layers. Currently, SiN x deposited by plasma enhanced chemical vapor deposition (PECVD) method has been the most widely used layer for AlGaN/GaN HEMT passivation [ 95 99 ]. The SiN x passivation is also essential i nmaintaining good inter device isolation blocking voltages for HEMT high voltage applications. An isolation blocking voltage of 540 V at a leakage current of 1 A/mm was achieved by combining uniform nitrogen isolation implantation with the use of carbon doped GaN buffer l ayers, with the structure passivated with a 400 nm PECVD SiN x layer [ 103 ]. However, there have been no reports on a systematic study of the

PAGE 74

74 effects of SiN x thickness, spacing between the contact window openings on the SiN x passivation layer, and the rf po wer used for the SiN x layer deposition on the isolation blocking voltage. In this work, we investigated the isolation current voltage ( I V ) characteristics between two Ohmic contact pads passivated with different thicknesses of SiN x passivation layer and the role of spacing between the contact window openings on the SiN x passivation layer. ATLAS/Blaze was used to model the isolation I V characteristics as a function of the dielectric thickness. The effects of rf power used for depositing the SiN x was al so studied. 4 2 2 Experimental AlGaN/GaN HEMT structures were grown on c plane sapphire by metal organic chemical vapor deposition (MOCVD) The epi layers consisted of a 1 m thick carbon doped GaN buffer layers followed by a 55 nm thick undoped GaN chann el layer, 20 nm of Al 0.25 Ga 0.75 N, and 2.5 nm GaN cap layer. The typical room temperature sheet resistance values ranged from 500 ohm/sq. to 550 ohm/sq. The HEMT fabrication was started with Ohmic contact deposition by lift off of e beam deposited Ti/Al/Ni /Au (25 nm/125 nm/45 nm/100 nm) multilayer followed by rapid thermal annealing (RTA) in a flowing nitrogen environment at 850C for 30 s ec A typical contact resistance of 0.6 mm was measured using the transmission line method (TLM). For the device is olation implantation, the Shipley Microposit STR 1045 positive photoresist was used as the mask to define the active region of the isolation testing sample and devices. All the samples were subjected to nitrogen ion implantation with implantation energies of 30, 160, and 400 keV, and the doses were 6 10 12 1.8 10 13 and 2.5 10 13 cm 2 respectively [ 103 ] The TRIM software was used to simulate the implantation profile. The implantation energies of 30, 160, and 400 keV were applied to the samples ste p by step during

PAGE 75

7 5 the nitrogen ion implantation, and the lower ion energy implantation of 30 keV was used to create deep traps near the wafer surface for isolating the two dimensional electron gas channel (2DEG). With these multiple energy and dose implant ations, a uniform vacancy density from the simulations of 5 10 20 cm 3 was created across a region 0.58 m thick. The metal contacts were passivated using SiN x by plasma enhanced chemical vapor deposition (PECVD) at 300C, followed by the wet etching of contact windows using Buffered Oxide Etch (BOE) solution (6 parts 40% NH 4 F and 1 part 49% HF). The isolation test structure used in this work consisted of 100 m 100 m square Ohmic contact pads separated by 1.7 and 4.7 m. Figure 4 7 illustrates a sch ematic cross section of an isolated tester for the AlGaN/GaN HEMT structures. The isolation leakage currents were measured with a Tektronix curve tracer 370A and a HP 4156 parameter analyzer. Finally, we also used ATLAS/Blaze (Automatically Tuned Linear Algebra Software) to simulate the electric field for the samples passivated with different thicknesses of SiN x passivation layer. 4 2 3 Results and Discussions Figure 4 8 shows the I V leakage characteristics across an isolation implanted region ranging f rom 1.7 to 4.7 m spacing, where the tester was passivated with different thicknesses, h, of the SiN x passivation layer. The distance, d, between the dielectric contact window openings was kept constant at 140 m. I V curves for two reference samples with out SiN x passivation layer are also included in Figure 4 8 The curves labeled with closed circles represent the unpassivated sample immersed in Fluorinert liquid during the I V measurements. The I V s for SiN x passivated samples were labeled with differen t symbols, as shown in Figure 4 8 If we define the isolation blocking voltage as the voltage at which the leakage current reached 10 A/mm, the blocking voltages of the testers immersed in Fluorinert solution and SiN x passivated testers with the SiN x l ayer thickness 375 nm increased from 200 V to 720 V, when

PAGE 76

76 the implanted gaps increased from 1.7 m to 4.7 m gaps. However, the corresponding field strengths across these gaps were very similar for both gaps, being around 1.2 1.5 MV/cm, close to the rep orted GaN field strength [ 106 ]. The Fluorinert solution is commonly used for device I V breakdown voltage measurements prior to device passivation [ 104,105 ]. Thus with proper passivation, the isolation blocking voltage depended on the breakdown strength of the semiconductor. Unlike the testers with 1.7 m implanted gap, the testers with 4.7 m implanted gad and without passivation layer or passivated with a thinner SiN x passivated layer exhibited an early breakdown voltage. Once the sample showed an ear ly breakdown voltage, it was permanently damaged and became an electrical short, as shown in Figure 4 9 In those samples not displaying such early breakdown voltage, the I V characteristics could be repeatedly measured. For the samples with 4.7 m impl anted spacer layer, the early breakdown voltage increased as the thickness of the passivation layer increased. Once the thickness of the SiN x passivation layer was larger than 375 nm, no early breakdown voltage was displayed. Therefore, besides the isola tion implanted spacing, the thickness of the SiN x passivation layer also affected the occurrence of the early breakdown voltage. The ATLAS simulated isolation blocking voltage as a function of the thickness of the SiN x passivation layer and the spacing of the isolation implanted region are illustrated in Figure 4 10 For the samples with 1.7 m of isolation implanted spacer layer, the simulated isolation blocking voltages were very consistent with the experimental results. Once the thickness of the SiN x passivation layer was larger than around 200 nm, the isolation blocking voltage became constant and dominated by the GaN breakdown strength. For the samples with 4.7 m of the isolation implanted spacer layer, due to the occurrence of the early breakdown voltage, there

PAGE 77

77 were some discrepancies between the simulated and experimental results for the samples with the thinner SiN x passivation layers. Similar to the samples with 1.7 m of the isolation implanted spacer layer, the isolation blocking voltage reac hed a constant value around 780 V for the samples passivated with the SiN x passivation layer thicker than 350 nm. This isolation blocking voltage was determined by the GaN breakdown strength. To take advantage of the wide energy bandgap GaN and AlGaN for high breakdown voltage device applications, it is important to have thick enough dielectric passivation to avoid the occurrence of the early breakdown. Beside the insufficient thickness of the SiN x passivation layer thicker causing the early breakdown vo ltage, the distance between the contact window openings on the SiN x passivation layer also impacted the incidence of the early breakdown voltage, as shown in Figure 4 1 1 For the samples were passivated with 375 nm SiN x passivation layer, an early breakdo wn voltage observed for the sample with a distance of 70 m between the SiN x contact window openings. The early breakdown voltage further reduced another 100 V for the sample with 35 m gap between the SiN x window openings. This was due to the surface br eakdown of the SiN x layer initiated by the electrons emitted from the metal dielectric junction under high field condition, often quoted as reaching flashover point [ 107 ]. The samples with the distance between the contact window openings larger than 100 m, there was no early breakdown observed. A similar trend of reducing the isolation blocking voltage for the samples passivated with 1.125 m SiN x layer is illustrated in Figure 4 1 2 An early breakdown voltage was shown for the samples with distances be tween the contact window openings of 35 or 75 m. Although the early breakdown voltages were slightly higher for the samples passivated with 1.125 m SiN x layer, the thicker passivation layer did not prevent the occurrence of the early breakdown. These r esults

PAGE 78

78 confirmed that the early breakdown voltage was also induced by the insufficient distance between the dielectric window openings. The effect of rf power during the SiN x passivation layer deposition on the isolation blocking voltage was also investig ated, as shown in Figure 4 1 3 Similar I V characteristics were obtained for the samples deposited with rf power ranging from 10 20 watts. There was a slight reduction of the isolation blocking voltage around 30 V for the samples with the SiN x passivatio n layer deposited at 30 watts of rf power. During the deposition, positive ions accelerated toward the substrate holder through a self bias voltage created by the rf power generated plasma and the magnitude of the self bias voltage was proportional to the level of the rf power. Once the self bias voltage reached a certain value, it introduced ion bombardment damage and increased the device leakage current. Figure 4 1 4 shows the typical drain I V characteristics of the AlGaN/GaN HEMTs with a gate dimensio n of 1 m 200 m and a gate to drain distance of the 37 m. The HEMTs were passivated with 375 nm of the SiN x layer deposited at 300C and employed an rf power of 15 W, and the distance between contact windows was kept above 150 m. An excellent 1000 V drain breakdown voltage and a saturation drain current of 310 mA/mm were achieved. Such devices are promising for high power and high breakdown voltage applications.

PAGE 79

79 Figure 4 1. Schematic cross section view of the AlN/GaN HEMT grown on sapphire. Sapphire GaN 2DEG 3.5 nm AlN S D G Thin Oxide

PAGE 80

80 Figure 4 2. EDS spectra of the AlN/GaN HEMT surface before and after the ozone treatment as well as the EDS spectrum of the AlN/GaN HEMT surface with an additional oxygen plasma exposure. Figure 4 3. Gate current of the AlN/GaN HEMTs treated with ozone (d mode HEMT) only and with an additional oxygen plasma exposure (e mode HEMT).

PAGE 81

81 Figure 4 4. DC IDS VDS characteristics of a n AlN/GaN HEMT with (left) no oxygen plasma exposure, (right) the gate area of the HEMT exposed to 24 sec of oxygen plasma. Figure 4 5. DC transfer characteristics for the d mode and e mode AlN/GaN HEMTs.

PAGE 82

82 Figure 4 6. G ate lag measurement data of the drain current as a function of pulse gate voltag e for the e mode and d mode HEMTs.

PAGE 83

83 Figure 4 7 Schematic cross sectional view of an isolation blocking voltage tester fabricated on the AlGaN/GaN HEMT structure. GaN buffer AlGaN Ohmic metal Ohmic metal sapphire substrate Implanted area h SiN x d 2DEG gap h: silicon nitride thickness d: distance between two openings gap= 1.7 or 4.7 m (between Ohmic metal pads)

PAGE 84

84 Figure 4 8 Isolation current voltage characteristics measured ac ross two 100 m 100 m Ohmic contact pads separated with an isolation implanted space ranging from 1.7 or 4.7 m. These testers were passivated with different thicknesses of SiNx passivation layer. The distance, d, between the dielectric openings was k ept at 140 m. Figure 4 9 Microscope pictures of device before (top) and after (bottom) experiencing early breakdown.

PAGE 85

85 Figure 4 10 Comparison between ATLAS simulated isolation blocking voltages and the experimental data, wher e the isolation blocking voltages were plotted as the functions of the thickness of the SiN x passivation layer and the spacing of the isolation implanted region. Figure 4 1 1 Isolation blocking voltage as a function of the distance between the two contact window openings on the samples passivated with 375 nm of SiN x layer.

PAGE 86

86 Figure 4 1 2 Isolation blocking voltage as a function of the distance between the two contact window openings on the samples passivated with 375 nm or 1.125 m of SiN x layer. The samples exhibited with an early breakdown voltage were labeled with stars and triangles. Figure 4 1 3 I V characteristics of testers passivated with 375 nm of SiN x layer deposited at different rf powers. The separation between the contact window openings was kept at 140 m and the implanted isolation gap was 4.7 m.

PAGE 87

87 Figure 4 1 4 Drain I V characteristics of the HEMT. The device exhibited a saturation current >310 mA/mm and a drain breakdown voltage o f 1000 V.

PAGE 88

88 CHAPTER 5 EFFECTS OF PROTON IRRADIATION ON HETEROJUNCTION TRANSISTORS 5 .1 Proton Irradiation Effects on Sb based Heterojunction Bipolar Transistors 5 .1.1 Background Quaternary In 0.37 Ga 0.63 As 0.89 Sb 0.11 based double heterojunction bipolar transis tors (DHBTs) have attracted a great deal of attention because of their high speed and ultra low turn on voltage compared to that of the conventional In AlAs /In 0.53 Ga 0.47 As single heterojunction bipolar transistor s [ 15 17 ] These characteristics are due to t heir smaller base band gap and favorable type I emitter base ( E B) junction and type II base collector (B C) junction T he higher valence band offset at the In 0.52 Al 0.48 As/In 0.39 Ga 0.61 As 0.77 Sb 0.23 E/B junction prevents the injection of holes from the base to the In 0.52 Al 0.48 As emitter, leading to higher electron injection efficiency. Furthermore, the type II band alignment of the In 0.39 Ga 0.61 As 0.77 Sb 0.23 /In 0.53 Ga 0.47 As ( B/C ) junction increases the current capability due to the postponed Kirk effect. Exce llent dc and rf performance of DHBTs with In 0.37 Ga 0.63 As 0.89 Sb 0.11 as the base layer have been demonstrated. Another area of interest is the response of these devices to ionizing radiation, since it is expected that their high speed, low power capabilitie s can be used in space borne applications such as satellite communication systems in addition to ground based military uses. Many papers have reported the radiation response of GaAs and InGaAs based HBTs, but no work has been done on the response of quate rnary Sb based HBTs [ 108 112 ]. Higher energy bandgap ( E g ) semiconductor devices are better for radiation hard. AlGaN / GaN high electron mobility transistors (GaN E g = 3.4 eV) irradiated with 17 MeV protons to a fluence of 7 10 13 protons / cm 2 barely exhib ited any degradation [113]. InGaP / GaAs HBTs (GaAs E g = 1.4 eV) showed some degradations with a similar dose of irradiation.

PAGE 89

89 In this study we report the effects of In 0.52 Al 0.48 As/In 0.39 Ga 0.61 As 0.77 Sb 0.23 /In 0.53 Ga 0.47 As DHBTs irradiated with 5 MeV protons from 2 10 11 to 2 10 15 protons/cm 2 E B and B C junction characteristics, emitter, base and collector sheet resistance, common emitter gain current and collector I V characteristics of the HBTs before and after irradiation were evaluated to analyze the damages in the crystal structure from high energy protons under various doses. 5 .1.2 Experimental The schematic of a double mesa In 0.52 Al 0.48 As/In 0.39 Ga 0.61 As 0.77 Sb 0.23 /In 0.53 Ga 0.47 As DHBT is illustrated in Figure 5 1. The epi layers were grown on Fe do ped semi insulating (100) InP substrates in a Riber 32P solid source MBE system equipped with arsenic, phosphorus and antimony valved cracker cells. Silicon and beryllium were the n type and p type dopants, respectively. The Sb 2 and As 4 flux w ere set to a beam equivalent pressure (B.E.P.) of 1.8 10 7 torr and 8.0 10 6 torr to control the antimony and arsenic composition respectively The DHBT layer structure consists of a 400 nm n type In 0.53 Ga 0.47 As subcollector doped to 2 10 19 cm 3 a 150 nm n typ e In 0.53 Ga 0.47 As collector doped to 1 10 16 cm 3 a 44 nm p type In x Ga 1 x As 1 y Sb y based layer Be doped with Be to 6 10 19 cm 3 a 50 nm n type InAlAs emitter doped to 5 10 17 cm 3 a 50 nm n type InAlAs doped to 8 10 18 cm 3 and a 50 nm n type In 0.53 G a 0.47 As emitter cap doped to 2 10 19 cm 3 The substrate temperature was 490 o C throughout the entire growth except the base layer. During the growth of the base layer, the substrate temperature was decreased to 450 o C to prevent beryllium out diffusion a nd Sb phase separation. mesa wet etch process described in detail previously. Pt/Ti/Pt /Au metal contacts were used for

PAGE 90

90 the base electrode and Ti/Pt/Au for the emitter and collector electrodes. The device dc parameters were measured with an Agilent 4156B parameter analyzer. Proton irradiations were performed at Korea Institute of Radiological & Medical Sciences (KIRAMS) by using Cyclotron. The proton energy at the exit of the cyclotron was 30 MeV. The proton energy at the sample was 5 MeV after the protons passed through aluminum degrader s The thickness of each aluminum degrader was 2.7 mm. The entire beam current was measured using a Faraday cup to calculate flux density 5 .1.3 Results and Discussions Figure 5 2 shows a microscope image of the double mesa D HBT (a) before and (b) after the proton irradiation and there were no visible degradation in appearance. However, there were changes obvious from the electrical measu rements, as shown in Figure 5 3. Figure 5 3 (top) shows the forward and reverse current voltage ( I V ) characteristics from the emitter base (E/B) junction as a function of the proton fluence. At the very low bias generation recombination region (<0.25 V) there was no increase in base current for HBTs irradiated with a proton fluence of 2 10 11 cm 2 however, there was a slight increase and a significant increase in base current for the fluences of 2 10 13 and 2 10 15 cm 2 respectively. The base curr ent at the reverse bias region exhibited a similar trend, and there was no increase in base current with low doses of the proton irradiation. The base currents were 1.5 and 3 orders higher biased at 0.5 V bias voltage for the fluences of 2 10 13 and 2 10 15 cm 2 proton irradiation respectively. The ideality factors of the E B junction were extracted from the base current with the E B junction biased at 0.25 V < V BE < 0.45 V. There was no obvious change in the ideality factor for the lower fluences of proton irradiations, as shown in Figure 5 3 (bottom), but the ideality factor increased from 1.2 to 1.9 for the HBTs with the fluence of 2 10 15 cm 2 This was consistent

PAGE 91

91 with the irradiation damage of the HBTs in the region dominated by the parasitic resistance ( V BE > 0.45 V). The parasitic resistance only increased significantly under the condition of the highest proton irradiation dose, in which the free carrier density was reduced due to the deep traps formation resulting from the high fluence ( 2 10 15 cm 2 ) proton irradiation. The forward and reverse current voltage ( I V ) characteristics and ideality factor from the base collector (B/C) junction are shown in Figure 5 4 as a function of the proton fluence. As compared to the E B junction, t he B C junction suffered a higher degree of damage, larger increase of collector current at both forward and reverse bias conditions as well as ideality factor observed due to more traps generated in the neutral base and collector regions. For the DHBTs that received the highest irradiation fluence (2 10 15 cm 2 ), the B C junction was seriously degraded and became an electrical short For the E B junction, we only needed to consider the intrinsic junction. However, due to the double mesa device layout used in this study, as shown in Figure 5 1, both the intrinsic junction under the emitter mesa and the extrinsic junction outside the emitter mesa needed to be considered for the BC junction. Besides DHBTs, there were also B C diodes without the emitter layer, fabricated on the same samples that showed I V characteristics similar to the B C of the DHBT (Figure 5 5 ). The B C diodes without the emitter mesa could be treated as the extrinsic B C junction. In order to study whether the intrinsic B C junction was also shorted, we needed to isolate the intrinsic B C junction from the extrinsic junction. This was achieved by measuring the IV characteristics of the E C junctions for the proton irradiated and reference samples by biasing the emitter and collector elec trodes while floating the base electrode. Since the base layer thickness was only 4 4 nm, much shorter than the distance of 5 m between the emitter mesa and the based metal electrode, the electrons injected by the emitter diffused through the thin base la yer and reached the collector in the

PAGE 92

92 intrinsic B C junction. Hence, it was reasonable to ignore any conduction from the extrinsic junction for the I V measurements between the E C junctions. Figure 5 5 shows the I V characteristics of E B, B C, and E C junctions of a reference DHBT (top) and DHBT irradiated with the fluence of 2 10 15 cm 2 (bottom). As discussed previously, the E B junction of the proton irradiated DHBT showed similar characteristics but with higher recombination leakage current, idea lity factor, parasitic resistance, and reverse bias leakage current. For the proton irradiated DHBTs, besides the shorted B C junction, the I V characteristics of the E C junctions were completely different from those of the reference DHBTs. The forwar d and reverse current of the biased E C junction for the reference DHBTs were clamped in both directions due to the n p n configuration of the E C junction. During the forward bias condition, the E B junction of the n p n structure between the E C electro des was turned on, but the B C junction in the n p n structure was in the reverse bias mode. Likewise, during the reverse bias condition, the B C junction of the n p n structure between E C electrodes was turned on, but the E B junction was under reverse bias. Thus for both reverse and forward bias conditions, the intrinsic E C junction consisted of a forward biased n p junction in series with a reverse biased p n junction. Therefore, both the forward and reverse current of the biased E C junction would b e clamped. However, the I V characteristics of the E C junction were identical to the I V characteristics of the E B junction for the proton irradiated DHBTs. The only possibility of having the same I V characteristics for both the E B and E C junctions was to have a shorted B C junction. Therefore, we could conclude that the intrinsic B C junction was also shorted for the DHBTs irradiated with the fluence of 2 10 15 cm 2 protons. The degradation mechanism was dominated by the displacement damage in th e base emitter and base collector

PAGE 93

93 junction space charge regions as well as the neutral base region gives rise to increased recombination currents in both E B and B C junctions [114]. The increase in the base current for the proton irradiated DHBTs also de graded the common emitter current gain. Figure 5 6 (top) shows the Gummel plots before and after the proton irradiation, while the bottom plot shows common emitter current as a function of collector current. As shown in the Figure 5 6 (top), the current gain decrease was not only due to the base current increase, but also to the decrease of collector current. The current gain of the reference sample peaked at a collector current of 0.1 A and it was reduced from 40 at 0.1 A to 20 at 10 7 A. There was aro und 10% and 40% reduction of current gain observed for the DHBTs exposed with the fluence of 2 10 11 and 2 10 13 cm 2 proton irradiation, respectively. For the DHBTs irradiated with the fluence of 2 10 15 cm 2 protons, there were significant base curre nt increase and collector current reduction. The peak of the current gain dropped from 40 to 1.6. The common emitter collector I V characteristics from the D HBTs before and after different proton dose irradiation s are shown in the Figure 5 7. Figure 5 7 (top) shows the overlapping between the common emitter collector I V characteristics of the reference DHBTs and the DHBTs irradiated with 2 10 11 cm 2 protons and the exposure with this fluence ( 5 MeV ) of protons is equivalent to around 4,500 years of exp osure in low earth orbit. There was only minimal degradation of increasing output conductance and decreasing e arly voltage due to the decrease of the effective base resistance resulting from the traps created by the proton irradiation. However, for the h igher fluences, more noticeable damage was observed as shown in Figure 5 7 (bottom). A larger increase of output conductance and lower e arly voltage were observed due to a significant decrease in the effective base resistance resulting from more traps cre ated by the proton irradiation In addition, t he re wa s a reduction in current change which was due to the

PAGE 94

94 increase in the base current that resulted from the excess recombination center created by the irradiation damage. The DHBTs irradiated with the fl uence of 2 10 15 cm 2 protons were not functional at all, as shown in Figure 5 7 (bottom). These results were consistent with the increases in the sheet resistance of the emitter, base, and collector layers as listed in Table 5 1 5 .2 Proton Irradiation Effects on AlN/GaN High Electron Mobility Transistors 5 .2.1 Background AlGaN/GaN high electron mobility transistors (HEMTs) are excellent candidates for RF/microwave power amplifiers because of their high power and high speed handling capabilities [ 73 82 ] AlN/GaN HEMTs using the large band gap of AlN (6.2 eV) as the barrier layer provide even better carrier confinement and exhibit lower gate leakage current, and the absence of alloy disorder (compared to AlGaN barriers) results in improvement of both low and high field carrier transport. A number of variations on AlN/GaN HEMTs have been explored with notable success [ 83 87 ] and show promise for high power, high temperature applications in telecommunications, hybrid electric vehicles, power flow control, and satellite communication systems for broadband data transmission and weather forecasting. For those space applications, it is necessary to measure the response of the electronic devices to the type of high energy proton (even up to a few hundreds MeV in the Van Allen belts) and gamma ray fluxes encountered in earth orbit [1 15 120 ]. For AlGaN/GaN HEMTs, several groups reported the effect of proton irradiation on the device performance, but there has been no work on AlN/GaN HEMTs. Luo et al. [121] rep orted the degradation of the electrical characteristics of unpassivated and Sc 2 O 3 passivated AlGaN/GaN HEMTs after 40 MeV proton irradiation at a dose of 5 10 9 cm 2 The drain source currents in the devices decreased by 15 20% at the higher fluence, whi le the extrinsic

PAGE 95

95 transconductance decreased by ~30% under the same conditions [ 121 ]. The drain current recovered 90% after a 10 min annealing at 300 C. However, both AlGaN/GaN and AlGaN/AlN/GaN HEMTs showed significant degradation, around 40% drain curre nt reduction, after the devices exposed to proton at a fluence of 10 14 cm 2 proton irradiated at 1.8 MeV [ 122,123 ]. Hu et al. [ 124 ] and Kim et al. [ 125 ] studied the energy dependence of proton induced degradation in AlGaN/GaN HEMTs up to 105 MeV; there wa s no significant degradation at the conditions of higher energy, >15 MeV, for the proton irradiations. In these cases, the damaged area was created in the substrate and not on the surface of the sample. In this paper, we report on the electrical changes in AlN/GaN HEMTs irradiated with 5 MeV protons at doses ranging from 2 10 11 to 2 10 15 protons/cm 2 The direct current, extrinsic transconductance, gate lag drain current, and gate characteristics of AlN/GaN HEMTs, as well as transmission line method b efore and after proton irradiation were used to study the effect of proton irradiation on HEMT performance. Low temperature annealing at 250 and 300 C was also used to anneal the proton irradiation damage. 5 .2.2 Experimental AlN/GaN HEMT structures were grown on c plane sapphire in a M olecular B eam E pitaxy (MBE) system equipped with an RF nitrogen plasma source from SVT Associates (SVTA). During growth, reflection high energy electron diffraction (RHEED) was used to monitor surface morphology. Other in situ measurements, including emissivity corrected surface temperature, thin film growth rate, and III/V flux ratio were performed by a combination of pyrometry and two color reflectometry (SVTA IS4000). The growth process started with surface nitridation at high temperatures using the rf nitrogen plasma source, followed by the growth of a thin AlN nucleation layer. Then, a 2 to 3 m low defect GaN buffer was grown. A

PAGE 96

96 relatively low TD density (< 110 8 cm 2 ) in these films has been previously confirmed by both etch pit density measurements, using atomic force microscopy (AFM), and by transmission electron microscopy (TEM) [ 126 ]. Finally, the AlN/GaN active layer was formed by growing a thin (<5 nm) strained AlN layer at about 700 C. A similar TD was obse rved in the AlN. A cross sectional TEM of AlN/GaN HEMT structure is illustrated in Figure 5 8 The optical and electrical properties were characterized by cathodoluminescence, Hall, and capacitance voltage measurements. The mobility and sheet carrier de nsity in the two dimensional electron gas channel ranged from 1300 1900 cm 2 /V s and 1.5 3.5 10 13 cm 2 respectively. The typical sheet resistance of the sample was below 200 / Prior to the normal device fabrication, the samples were treated with an UV generated o zone for 1 min to protect the AlN surface layer. The ozone treated surface exhibited resistance to positive resist developer solution, which is known to easily attack the AlN. Device fabrication began with mesa isolation by Cl 2 /Ar indu ctively coupled plasma (ICP) etching (150 W source power, 40 W RF chuck power). Isolation currents were less than 2 A at 40 V bias for a mesa depth of 1 2 00 Ohmic contacts were formed by lift off of e beam deposited Ti/Al/ Ni /Au based metallization, an d subsequently annealed at 8 50C for 30 s ec under a N 2 ambient. The gate fingers with a dimension of 40 0 nm 200 m were defined using a Raith electron beam direct write system. The distance between the source and drain was 2 m. A schematic cross section view of the device is shown in Figure 5 9 (top). The dc characteristics of the HEMTs were measured with a Tektronix curve tracer 370A and an HP 41 56 parameter analyzer For each sample, five devices were measured. For the gate lag measurements, the gate voltage, V G was pulsed from 5 V to different voltages a pulse generator at different frequencies with a 10% duty cycle and the drain current was monitored with an oscilloscope.

PAGE 97

97 Proton irradiations were performed with a MC 50 (Scanditronix) cyclotron located at the Korea Institute of Radiological & Medical Sciences (KIRAMS). The proton energy at the exit of the cyclotron was 30 MeV and the proton energy at the sample was reduced to 5 MeV after protons passed aluminum de grader s The thickness of each aluminum degrader was 2.7 mm. The entire beam current was measured using Faraday cup to calculate flux density. The stopping depth of 5 MeV protons in the AlN/GaN substrate was more than 100 m, estimated with the TRIM prog ram. 5 .2.3 Results and Discussions Figure 5 9 (bottom) shows microscope image of the AlN/GaN HEMTs (a) before and (b) after the proton irradiation There was no visible degradation of the metallization for both Ohmic and gate contacts. However, there were changes in the dc performance, as shown in Figure 5 10 Figure 5 10 ( left ) shows the typical drain I V characteristics of the AlN/GaN HEMT before and after proton irradiation with a fluence of 2 10 11 cm 2 and Figure 5 10 (right ) shows the typical d rain I V characteristics of the AlN/GaN HEMT exposed to the proton irradiations with fluences of 2 10 13 and 2 10 15 cm 2 The drain current decreased slightly as the drain voltage increased and this decrease was due to self heating effect. After expo sure to the proton irradiation, the saturation drain current decreased, and source drain resistance (the reciprocal of the slope for the drain I V in the low drain voltage region) increased substantially. 10% changes of the saturation drain current and th e source drain resistance for the HEMT exposed to the irradiations with fluences of 2 10 11 were observed. For samples exposed to the higher fluences, 2 10 13 and 2 10 15 cm 2 the degradation of the drain current and source drain resistance were 15 to 35%, respectively. High energy incident protons damage the crystal lattice and displace atoms to create charged defect (trap) centers. With more defect centers, carrier

PAGE 98

98 scattering and carrier removal become more severe and the electron concentration in the channel and saturated carrier velocity were reduced. These results were consistent with the increases in the sheet resistance, transfer resistance, and contact resistivity of the Ohmic contact measured with transmission line method listed in Table 5 2 The data of the sheet resistance and transfer resistance listed in Table 5 2 were averaged from 8 TLMs. By using TLM to study the proton irradiation damage, we could separate the effect of gate contact degradation on the drain current reduction. Bes ides the saturation drain current reduction after the proton irradiation, the drain current of the AlN/GaN HEMT also became negative in the low drain bias voltage range (0 0.5 V), as illustrated in Figure 5 10 This was due to gate leakage current when th e gate was biased at higher positive gate bias voltages. It is worth noting that the gate voltage was modulated from 3 V to 3 V with 1 V steps as shown in Figure 5 10 which was higher than the 1 to 2 V for a typical Schottky gate on AlGaN/GaN HEMTs. A s illustrated in Figure 5 11 no Schottky diode behavior was observed. The gate current of the AlN / GaN HEMT behaved as a leaky metal oxide semiconductor (MOS) diode like I V This MOS diode like behavior was due to the surface of the AlN layer been oxidiz ed by the ozone treatment prior to the device fabrication. Although this gate oxide layer could be modulated the drain current of the pre irradiated HEMT from +3 V to 4 V of the gate voltage, which suffered relat ively high gate leakage current, t his oxid e layer apparently was damaged during the proton irradiation and produced gate leakage current, and both the forward and reverse gate currents of the AlN/GaN HEMT increased significantly, causing the drain current of the AlN/GaN HEMT to become negative in the low drain bias voltage range. It has been previously reported that proton induced displacement of nitrogen atoms can be repaired through room temperature annealing, but displacement of gallium atoms

PAGE 99

99 from the crystal lattice is rarely repaired and will play a primary role to form charged defect centers [ 127 ]. We annealed these samples at 250, 300, and 350 C from 5 minutes to 3 hours. The forward and reverse gate current decreased 10% as shown in Figure 5 11 but there was no obvious drain current impr ovement obtained. Figure 5 12 (top) shows the normalized extrinsic transconductance and square root of the normalized drain current measured at a fixed drain voltage of 5 V and the threshold voltage was extracted from the intercept of square root of the normalized drain current to the gate voltage. Five devices were measured and the average of maximum extrinsic transconductance, g m and threshold voltages, V th were plotted as a function of the proton irradiation fluence as shown in Figure 5 12 (bottom). The g m of the pre irradiation HEMT was 287 mS/mm, and 17%, 24% and 35% reductions of the g m were obtained after proton irradiation with the fluences of 2 10 11 2 10 13 and 2 10 15 cm 2 respectively. The threshold voltage shifted positively from 2.14 V for the pre irradiation HEMT to 1.97 V after irradiation with protons at the fluence of 2 10 15 cm 2 The proton induced defect centers led to decreases of carrier density and mobility, and thus reduced maximum transconductance and resulted in the V t h shift with the increasing proton dose. A typical gate lag data of the drain current as a function of pulse gate voltage for the AlN / GaN HEMTs before and after the proton irradiation are shown in Figure 5 13 The gate lag measurements were conducted a t a f ixed drain voltage of 5 V, and the gate voltage was pulsed lag has been attributed to the influence of surface states, which is an excellent method to identify surface traps. For the pre irradiation HEMT, there was no dispersion of the drain current observed, regardless of the different frequency of the gate pulses. This indicates that the ozone treatment prior to the device fabrication provided an effective p assivation effect and did not

PAGE 100

100 introduce defects on the AlN / GaN HEMT surface. For the proton irradiated HEMTs, there was no drain current reduction for the gate lag measurements at 1 kHz. There was almost no decrease of the drain current for the 2 10 11 cm 2 proton irradiated HEMT at 100 kHz, and 4.3% and 9.2% drain current reduction during the gate lag pulse measurement performed at 100 kHz for 2 10 13 and 2 10 15 cm 2 proton irradiated HEMTs, respectively. The degradations observed in the gate lag mea surement were much less severe than those obtained for the drain current and g m This confirmed that minimal concentrations of surface traps were created by the high energy proton irradiation, which is consistent with the published results [ 122 125 ]. Ta ble 5 1. Emitter, base and collector Sheet resistance before and after proton irradiation Resistance ( ) Before irradiation 2 10 11 cm 2 2 10 13 cm 2 2 10 15 cm 2 Emitter Resistance 41 44 53 61 Base Resistance 1100 1183 1282 1400 Collector Resistance 10 11 13 15 Table 5 2 Sheet, contact and transfer resistance before and after proton irradiation Resistance Pre irradiation 2 10 11 cm 2 2 10 13 cm 2 2 10 15 cm 2 Sheet Resistance( ) 213 2.6 259 4.3 263 5.5 271 4.1 Contact Resistance( cm 2 ) 9.2 0.15 1 0 5 2.2 0.04 10 4 2.8 0.06 10 4 3.6 0.07 10 4 Transfer Resistance( mm) 1.4 0.04 2.39 0.07 2.71 0.09 3.12 0.07

PAGE 101

101 Figure 5 1. Schematic of a double mesa In 0.52 Al 0.48 As/In 0.39 Ga 0.61 As 0.77 Sb 0.23 /In 0.53 Ga 0.47 As DHBT. Figure 5 2. Optica l microscope image of the double mesa D HBT (a) before and (b) after the proton irradiation

PAGE 102

102 Figure 5 3. E B junction forward I V characteristics (top) and ideality factor (bottom) as a function of the proton dose.

PAGE 103

103 Figure 5 4. B C junction I V characteristics (top) and ideality factor (bottom) as a function of the proton dose.

PAGE 104

104 Figure 5 5. I V characteristics of E B, B C, and E C junctions of a reference DHBT (top) and DHBT irradiated with the fluence of 2 10 15 cm 2 (bottom).

PAGE 105

105 Figure 5 6. (top) Gummel plots before the proton irradiation and after different proton dose irradiation. (bottom) DC current gains as a function of the collector currents

PAGE 106

106 Figure 5 7. Common emitter characteristics from the D HBTs before and after different proton dose irradiation.

PAGE 107

107 Figure 5 8 Cross sectional TEM of AlN/GaN HEMT structure.

PAGE 108

108 G S D G S D (a) (b) Figure 5 9 (top) Schematic cross secti onal diagram of the AlN/GaN HEMT device and (bottom) optical microscope images of (a) pre and (b) post irradiated HEMT devices

PAGE 109

109 Figure 5 10 ( left ) I DS V DS characteristics of AlN/GaN HEMTs before and after 2 10 11 protons/cm 2 irradiatio n. (right ) I DS V DS characteristics of AlN/GaN HEMTs exposed to 2 10 13 and 2 10 15 protons/cm 2 irradiation. Figure 5 11 Gate current of AlN/GaN HEMTs before and after proton irradiation, and after subsequent thermal annealing

PAGE 110

110 Figure 5 12 (top) DC transfer characteristics for pre and post proton irradiated AlN/GaN HEMT devices, and (bottom) extracted threshold voltage and extrinsic maximum transconductance as a function of proton fluence (2 10 11 2 10 15 pro tons/cm 2 )

PAGE 111

111 Figure 5 13 Gate lag measurement data for the drain current as a function of the pulsed gate voltage

PAGE 112

112 CHAPTER 6 NI AND PT GATED ALGAN/GAN ELECTRON MOBILITY TRANSISTORS 6 .1 Background AlGaN/GaN High Electron Mobility Transist ors (HEMTs) show great promise for high power and high frequency operation [ 128,129 ] and for applications such as high frequency wireless base stations and broad band links, commercial and military radar and satellite communications [ 130 132 ]. In recent r eports, hot electron induced degradation has been observed in undoped AlGaN/GaN HEMTs on sapphire, SiC and Si substrate during DC and RF stresses [ 133 135 ]. Several degradation mechanisms that suppress device performance and reliability have been reported ranging from hot electron induced trap generation to field driven mechanisms [ 136 141 ]. Ni/Au metallization has been widely employed as the gate contact for AlGaN/GaN HEMT and consequently it has been used in most HEMT reliability studies to date. Pt based metal schemes have been used as p Ohmic contacts on p type GaAs to improve the reliability of AlGaAs/GaAs heterojuction bipolar transistors [ 142,143 ]. By alloying at low temperature, Pt diffused into GaAs and reacted with GaAs to form a thin PtAs 2 layer to achieve low contact resistance and excellent device stability. Pt based metallization in the form of Pt/Ti/Pt/Au was also employed as the gate contacts for InGaP/GaAs, InAlAs/InGaAs [ 144,145 ], and AlGaAs/InGaAs HEMT to improve the device reliabil ity, and an intentional annealing step at 250 to 350C was used to sink the Pt into the gate contact semiconductor layer to adjust the threshold voltage of the HEMTs [ 146 ]. This allowed both depletion and enhancement mode mode HEMTs to be fabricated on th e same wafer. Pt/Ti/Au metallizaton has also been used as the gate contacts on GaN and AlGaN. The thermal stability of Pt/Ti/Au was much higher as compared to Ni/Au gate contacts [ 147 ]. The thickness of the Ti layer in the Pt/Ti/Au played an important r ole in the resulting gate leakage current upon thermal stress [ 148 ].

PAGE 113

113 In this work we compared the critical voltages for off state stress, Schottky gate forward and reverse current characteristics, as well as drain I V characteristics for HEMTs fabricate d with the Ni/Au and Pt/Ti/Au gate metallization. X ray photoelectron Spectroscopy (XPS) was used to examine the thermal stability of the Ni and Pt metal contacts deposited on GaN. 6 .2 Improvement of Off State Stress Critical Voltage by Using Pt gated AlGa N/GaN High Electron Mobility Transistors 6 .2.1 Experimental AlGaN/GaN HEMT heterostructures were grown on c plane sapphire by metal organic chemical vapor deposition (MOCVD). The epi layers consisted of a 1 m thick carbon doped GaN buffer layer followed by a 55 nm thick undoped GaN channel layer, 21 nm of Al 0.25 Ga 0.75 N, and 2.2 nm GaN cap layer. The HEMT fabrication was started with Ohmic contact deposition by lift off of e beam deposited Ti/Al/Ni/Au multilayer followed by rapid thermal annealing (RTA) a t 850C for 30 s ec in a nitrogen environment. A typical contact resistance of 0.6 mm was measured using the transmission line method (TLM). For the device isolation, multiple doses and energies of nitrogen ion implantation were used to maintain a planar geometry in the fabricated device and reduce parasitic leakage current. Shipley Microposit STR 1045 positive photoresist was employed to protect the active region of the devices Ni/Au (20 nm/80 nm) and Pt/Ti/Au (10 nm/20 nm/80 nm) based Schottky gate me tallization was defined by optical lithography and followed with standard lift off of the e beam deposited metals. The gate dimension was 1 m 200 m and the distances of gate to source (L gs ) and gate to drain (L gd ) were 1 m and 3 m, respectively. Th e HEMTs were passivated using 400 nm of SiNx with a plasma enhanced chemical vapor deposition (PECVD) system at 300C. The metal windows were opened with Buffered HF solution. The device I V characteristics were measured with a Tektronix curve tracer 370 A and an HP 4156 parameter analyzer. The HEMTs off state step

PAGE 114

114 stresses were performed in the dark at room temperature with the gate biased up to 90V reverse gate voltage at a fixed source drain bias of 5 V using an HP 4156C semiconductor parameter analyze r. XPS was used to examine the thermal stability of the Ni and Pt deposited on GaN. The system employed a Mg anode for the x ray source; take off angle was kept at 45 for all measurements, and the pass energy was 35.75 eV. Metal films of 1 nm Ni or P t were deposited on 3 m GaN and half of the samples were annealed in vacuum chamber (1 10 9 t orr) at 300C for 30 minutes. Prior to the metal deposition, the GaN samples were rinsed in 1:1 HCl:H 2 O solution for 3 minutes, exposed to UV ozone for 25 minu tes, dipped in buffered HF for 5 minutes and rinsed in de ionized water. The samples were then loaded into a III N MBE system for a thermal clean at 700C for 30 minutes in the UHV environment. This cleaning process was found to be efficient at reducing the native oxide content and surface carbon. The GaN samples were removed from the UHV system and immediately loaded into the e beam evaporator for Ni or Pt deposition. This limited the exposure of the GaN to atmosphere to maintain the native oxide thic kness and contamination to a minimum. Post metal deposition, the samples were immediately loaded into the XPS system to minimize exposure to atmosphere. 6 .2.2 Results and Discussions 15 separate HEMTs with Ni/Au or Pt/Ti/Au gate metallization were stress ed for 60 seconds at each gate voltage step, while grounding the source electrode and maintaining +5 V on the drain. The stress started at 10 V of gate voltage and the voltage step was kept at 1 V. During the step stress, besides monitoring I g gate to source leakage current, I gs and gate to drain leakage current, I gd were also measured. Between each step stress, the drain I V characteristic, 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 all recorded. Self heating

PAGE 115

115 effects were negligible based on the low drain source currents under our test conditions, a fact supported by thermal simulations. The critical voltage of the off state step stress was d efined as the onset of I g increase during the stress. The critical voltages of 15 Ni/Au gated HEMTs were varied from 45 to 65 V. As shown in Figure 6 1, the HEMTs with Ni/Au gate metallization exhibited a critical voltage around 55 V. The I g suddenly increased and reached a compliance of 1 mA/mm, which was set to protect gate. It has been reported that once the gate voltage passed the critical voltage, a decrease of saturation drain current and increases of the source and drain resistance appeared [ 1 39 ]. Similar degradations were observed for the stressed Ni/Au gated HEMTs used here. However, there was no critical voltage observed for the HEMT with the Pt/Ti/Au gate metallization up to 100V, which was limited by the instrument used in this experi ment. This suggested that the use of Pt based gate metallization could extend the operating bias conditions and improve the device reliability. The Schottky barrier height and ideality of the Pt/Ti/Au were 1.23 eV and 1.21, respectively, which did not e xhibit noticeable changes as a result of the bias stressing, as illustrated in Figure 6 2. On the contrary, the HEMTs with Ni/Au gate metallization showed significantly higher gate reverse bias leakage current and much lower breakdown voltage. The forwar d gate characteristics of the Ni/Au gate contact appeared very leaky after the stress and the Schottky height reduced from 1.09 V to 0.66 V after stress. The lower Schottky barrier height of Ni due to the lower work function of Ni, as compared to Pt, resu lted in almost an one order higher gate leakage current at the gate bias voltage, Vg, > 60 V. The gate leakage was due to the emission of electrons from a trapped state near the metal semiconductor interface into a continuum of states which associated wi th each conductive dislocation and added the

PAGE 116

116 reference [ 149 ]. The excessive leakage current at the higher reverse biased Schottky gate contact was identified as a key indicator of poor device reliability. Besides the gate I V characteristics, the drain I V of Pt/Ti/Au gated HEMTs also exhibited a minimal change after the stress (around 2% reduction) as compared to the 12.5% reduction for the Ni/Au gate HEMTs as shown in Figure 6 3 due to the increase of source resistance and drain resistance after the re verse gate voltage passed the critical voltage [ 139 ]. The threshold voltage of the Pt gate HEMTs was around 2.3 to 2.4 V, which slightly shifted to the positive side as compared to 2.7 to 2.8 V for the Ni gated HEMT, due higher Schottky barrier height of the Pt gated HEMTs. There was no changes of the threshold voltage were observed for the Pt gated HEMTs. However, the threshold voltage of the Ni gated HEMTs shifted to 2.4 V. This positive threshold voltage shift of the Ni gate after stress was at tributed to gate metal interacted with the nitride, which was reported by several groups [ 150 151 ]. These result were consistent with the gate I V and off state stress data that the reliability of the HEMTs was improved by using Pt/Ti/Au gate metallizatio n. For As based HEMTs and HBTs, Pt based metallization is usually annealed at 250 350C for 30 seconds to improve the gate contact stability and this process was also used to shift the threshold voltage of the HEMT for fabricating enhancement mode de vices, as mentioned previously [ 144 146 ]. The Pt/Ti/Au gate contacts were heated up at 300C for 60 mins during the silicon nitride deposition. In order to study the interaction between the Pt and GaN after 300C annealing, special XPS samples with 1 nm Ni and Pt deposited on the GaN were prepared. The metal films were intentionally kept thin so the metal/GaN interface could be studied in XPS. As shown in Figure 6 4 ( a and b ), the spectra of Pt 4f XPS spectra for the as deposit Pt/GaN sample and post an nealed sample, there were no changes observed in the spectra.

PAGE 117

117 This result confirmed the off state stress results that the Pt/Ti/Au gate contacts were stable after electrical and thermal stress; no observable chemical reaction at the interface had taken pl ace. By contrast, as shown in Figure 6 4 ( c and d ), by comparing the pre and post anneal spectra, the Ni oxide peak increased and the Ni metal peak decreased significantly. The discrepancy of the Ni O peak position for the as deposit and annealed samples was due to charging effect. These results indicated the Ni oxidized during the annealing in the vacuum chamber. By observing the O 1s peak, as shown in Figure 4 (e) and (f), and its changes between pre and post anneal, it was determined that the O Ga bo nding portion decreased and the O Ni bonding portion increased. This led to the conclusion that the Ni film stripped oxygen from the native oxidized GaN surface and created a layer of NiO. The reaction between the Ni and native oxide could also happen d uring the on state stress reported by Singhai et al [ 152 ]. The Ni O formation was just one the indications for the device degradation. It has been reported that the defects generated under the gate metal and pit formation along the gate edge close to the drain contact side [ 153 ]. All these defects would affect the gate leakage current, drain current and the efficiency of the gate to modulate the channel. Figure 6 1 Off state gate currents as a function of gate for the HEMTs fabricated with Ni /Au or PT/Ti/Au gate metallization.

PAGE 118

118 Figure 6 2 Forward (top) and reverse (bottom) Schottky gate characteristics before and after the off state stress for the HEMTs fabricated with Ni/Au and Pt/Ti/Au gate metallization.

PAGE 119

119 Figure 6 3 Drain I V characteristics of HEMTs before and after the off state stress for the HEMTs fabricated with Ni/Au and Pt/Ti/Au gate metallization.

PAGE 120

120 Figure 6 4 XPS spectra of (a) Pt 4f XPS spectra for as dep osit Pt/GaN sample. (b) Pt 4f XPS spectra for the Pt/GaN sample annealed at 300C for 30 mins. (c) Ni 2p XPS spectra for as deposit Ni/GaN sample. (d) Ni 2p XPS spectra for the Ni/GaN sample annealed at 300C for 30 mins. (e) O 1s XPS spectra for as dep osit Ni/GaN sample. (f) O 1s XPS spectra for the Ni/GaN sample annealed at 300C for 30 mins.

PAGE 121

121 CHAPTER 7 PT GATED ALGAN/GAN HIGH ELECTRON MOBILITY TRANSISTOR BASED SENSORS 7 .1 Background Hydrogen is a colorless, tasteless and flammable gas subject to co mbustion and explosion risk at concentrations between 4% and 75%. Detection of hydrogen leaks at room temperature in various environments has been gaining much interest as applications emerge in hydrogen powered automobiles, proton exchange membrane fuel cells, solid oxide fuel cells for spacecraft, and other industrial long term sensing applications. The advantages of using sensors fabricated with wide bandgap semiconductor materials include the ability to operate the sensors at elevated temperatures and to be integrated with on chip wireless data transmission circuits [ 154 161 ]. Specifically, the AlGaN/GaN high electron mobility transistor (HEMT) structure is particularly well suited for these applications. Most of the hydrogen sensing studies to dat e have been conducted with hydrogen balanced with dry nitrogen and a few experiments were performed in dry air conditions. However, in the real applications for detecting hydrogen leaks, humidity may play a significant role on the hydrogen sensing and the humidity in the major cities in the US are often quite high, >50% [ 162 ]. Therefore it is important to study the effect of humidity on the AlGaN/GaN HEMT based hydrogen sensors. In this work, we report on the electrical characteristics of AlGaN/GaN HEMT based Schottky diode exposed to hydrogen containing air ambient with and without relative humidity (RH) under reverse biased conditions and we propose a sensing mechanism to explain the decreasing sensitivity under increasing RH.

PAGE 122

122 7 .2 Effect of Humidity o n Hydrogen Sensitivity of Pt Gated AlGaN/GaN High Electron Mobility Transistor Based Sensors 7 .2.1 Experimental HEMT layer structures were grown on c plane Al 2 O 3 substrates by molecular beam epitaxy (MBE). The layer structure included an initial 2 m thi ck undoped GaN buffer followed by a 35 nm thick unintentionally doped Al 0.28 Ga 0.72 N layer. The sheet carrier concentration was ~ 1 10 13 cm 2 with a mobility of 1080 cm 2 /V s at room temperature. Mesa isolation was achieved by using an inductively coupled plasma system with Ar/Cl 2 based discharges. The Ohmic contacts were formed by lift off of Ti (200 )/Al (1000 )/TiB 2 (200 )/Ti (200 ) /Au (800 ). The metals were deposited by Ar plasma assisted RF sputtering at pressures of 15 40 mTorr and rf (13.56 MHz) powers of 200 250 W. The contacts were annealed at 850 C for 45 s under a flowing N 2 ambient in a Heatpulse 610T system. A 100 t hick Pt Schottky contact was deposited by e beam evaporation for the Schottky metal. This is needed for making the device sensitive to hydrogen through catalytic dissociation of molecular hydrogen. The final step was deposition of e beam evaporated Ti/Au (300 /1200 ) interconnection contacts. The individual devices were diced and wire bonded to carriers. The gas sensing experiments were performed in a tube furnace that contained electrical feed through connected to either an HP4145 parameter analy zer or an I V measurement system as shown in Figure 7 1. The hydrogen concentration used in this study was fixed at 1% H 2 which was diluted with dry or wet mix from a hydrogen source of 4% H 2 balanced with nitrogen. Mass flow controllers were used to co ntrol the gas flows through the chamber, and the devices were exposed to either dry air, humid air, or 1% H 2 mixed with air. The humidity of the gas stream was controlled by the ratio of dry air to humid air introduced in the test chamber. The

PAGE 123

123 relative h umidity of the gas stream was monitored with a humidity meter mounted in the downstream section of the test chamber. 7 .2.2 Results and Discussions Figure 7 2 illustrates the hydrogen sensing sensitivity of the HEMT diode before and after exposure of 1% H 2 mixed with dry air to the diode with the diode biased from 2 to 0.7 V. For these diodes, the current increases upon introduction of H 2 through a lowering of the effective barrier height [ 163,164 ]. The changes of the diode current before and after the hy drogen exposure were normalized to the diode current before the hydrogen exposure. The sensitivity was higher for the diode in reverse biased conditions. The diode current was orders smaller in the reverse bias region ( 2 to 0 V) as compared to the forwa rd bias conditions (0 to 0.7 V). Thus, any small change of the diode current due to exposure of hydrogen in the reverse bias conditions would result in larger normalized current change ( I/I) and the higher sensing sensitivity in the reverse bias conditio n has been reported by several groups [ 165 167 ]. Since the reverse bias condition offered better sensitivity, it was used to study the effect of humidity on the hydrogen sensing sensitivity in this work. Figure 7 3 (top) shows the time dependence of th e diode current for a HEMT sensor biased at a 1.5 V and exposed 1% H 2 balanced with air with different relative humidity. The 1% H 2 mixture was switched into the chamber through a mass flow controller for 180 s ec and then switched back to the humid air. Both sensing response and recovery times of the sensor were less than 1min and the sensor also demonstrated good recyclability. However, the sensitivity for 1% H 2 linearly decreased as the relative humidity increased, as illustrated in Figure 7 3 (bottom ). Both relative humidity and oxygen partial pressure have been reported to degrade the electrocatalytic activity of platinum [ 168 172 ]. The absorbed water vapor and

PAGE 124

124 oxygen blocked available surface adsorption sites of platinum for H 2 and lowered the co ncentration of hydrogen at the metal/semiconductor interface. As shown in Figure 7 3, the sensor exhibited a fast response for the recovery time, when the gas ambient switched from hydrogen containing humid air to non hydrogen containing humid air. How ever, a slow recovery time was observed, when the gas ambient switched from hydrogen containing humid air to non hydrogen containing dry air, as illustrated in Figure 7 4. This slow recovery behavior was similar to the results obtained in our previous stu dy conducted using dry nitrogen as the background ambient, which showed a long and slow recovery time [ 173 174 ]. The slow recovery time in that work was, we now believe, mistakenly attributed to longer times required to purge the residue hydrogen out of t he gas chamber. In this work we note that the presence of water molecules in the ambient shortened the recovery time. It has been reported that Pd or Pt adsorbs H 2 O and catalytically dissociates the adsorbed H 2 O to form OH molecules with assistance of su rface chemisorbed atomic oxygen [ 175,176 ]. The atomic H would readily react with OH to form H 2 O, thus the presence of humidity consumed the atomic hydrogen and reduced the sensor recovery time. In conclusion, AlGaN/GaN HEMT diodes were used for hydrog en detection and the effect of humidity on hydrogen sensing sensitivity was investigated. The absorbed water and oxygen molecules blocked available surface adsorption sites of platinum for H 2 absorption, and lowered the concentration of atomic hydrogen at the metal/semiconductor interface. However, the humidity enhanced the efficiency of the recovery after the sensor was exposed to the hydrogen containing ambient.

PAGE 125

125 Figure 7 1 Schematic of the hydrogen sensing system. Figure 7 2 Sen sing sensitivity, defined as the ratio of diode current change to the diode current in 1% hydrogen balance with air as a function of the AlGaN/GaN HEMT diode bias voltage.

PAGE 126

126 Figure 7 3 (top) Time dependence of absolute value of the AlGaN/GaN HEMT diode current biased at 1.5 V as the gas ambient switched back and forth between 1% hydrogen balance and air with different humidity. (bottom) Hydrogen sensitivity as a function of the humidity.

PAGE 127

127 Figure 7 4 Time dependenc e of recovery time characteristics of the AlGaN/GaN HEMT diode current biased at 1.5 V as the gas ambient switched back and forth between 1% hydrogen balance and air with 100% humidity as well as dry air.

PAGE 128

128 CHAPTER 8 SUMMA RY 8 .1 InAlAs/InGaAsSb/InGaAs D ouble Hetrojunction Bipolar Transistor s High speed In 0.52 Al 0.48 As/In 0. 42 Ga 0. 58 As 0.7 7 Sb 0. 23 /In 0.53 Ga 0.47 As DHBTs have been fabricated employing trilevel resist system to promise the submicron emitter finger definition and using self aligned processing to n arrow electrode separation and minimize parasitic resistances and capacitances. Guard ring s w ere formed to preserve the current gain at small emitter dimension for improving dc and rf performance. DHBTs with InGaAsSb base exhibited a low turn on voltage with a high dc current gain A peak f T of 268 GHz and a f max over 450 GHz of the InGaAsSB based DHBT with an emitter size of 0.65 8.65 m 2 were achieved. The submircon InGaAsSb DHBTs demonstrated an excellent potential to be used for high speed perform ance. 8 .2 InAlN /GaN High Electron Mobility Transistors The effect s of annealing on the mobility and sheet carrier concentration of InAlN/GaN HEMTs were studied through the Hall measurement and TLM. An optimal annealing temperature of 800C provided a min imum transfer resistance of 0.65 ohm mm and aspecific contact resistivity of 2 10 5 ohm cm 2 The InAlN/GaN HEMTs without cap showed a higher maximum drain current of 1.3 A/mm and a maximum extrinsic transconductance of 366 mS/mm. InAlN/GaN HEMTs with a thin GaN cap layer showed a lower gate leakage current. 8 .3 Passivation on AlN/GaN and AlGaN/GaN High Electron Mobility Transistors T he effect s of ozone and oxygen plasma treatment on AlN/GaN HEMT gate diodes and drain I V characteristics w ere investigat ed. Ozone treatment of AlN/GaN HEMT produced an oxide layer which effectively passivated and protected the AlN surface to chemically resist the attack of AlN by the AZ positive photoresist developer used for subsequent device fabrication.

PAGE 129

129 With an additio nal oxygen plasma treatment on the gate area prior to the gate metal deposition, enhancement mode AlN/GaN HEMTs were realized and t he gate characteristics of the oxygen plasma treated HEMT did not show Schottky gate forward turn on characteristics; instead MOS gate characteristics were observed. The EDS scans confirmed the oxide formation on oxygen plasma treated AlN/GaN HEMT surface. The effects of the SiN x layer thickness, the distance between the dielectric window openings, the distance between the Oh mic contact pads and rf power used for SiN x layer deposition on the isolation blocking voltage were studied. Below a certain critical thickness of SiN x and using insufficient distance between the contact window openings, an early breakdown of the isolation blocking voltage was observed and the devices suffered permanent damage due to surface breakdown phenomenon. The isolation blocking voltage displayed a slight degradation for sample passivated with the SiN x layer deposited with an rf power of 30 W. HEMTs with a gate dimension of 1 m 200 m and passivated with the optimized SiN x passivation conditions showed a drain breakdown voltage of 1000 V and a saturation drain current of 310 mA/mm. The roughness of the annealed Ni/Al/Pt/Au ohmic conta ct surface and edge could also have significant impacts on the breakdown voltage and will be addressed in the future work. 8 .4 Proton Irradiation on InAlAs/InGaAsSb/InGaAs Double Hetrojunction Bipolar AlN/GaN Transistor s and High Electron Mobility Transist ors Both of In 0.52 Al 0.48 As/In 0.39 Ga 0.61 As 0.77 Sb 0.23 /In 0.53 Ga 0.47 As DHBTs and AlN/GaN HEMTs were irradiated with 5 MeV protons at fluences of 2 10 11 2 10 13 and 2 10 15 protons/cm 2 and showed a remarkable resistance to proton irradiation damage. The InAlAs/InGaAsSb/InGaAs DHBTs irradiated with a dose of 2 10 11 cm 2 which was equivalent to around 40 years of exposure in low earth orbit showed minimal changes in the junction

PAGE 130

130 ideality factor, generation recombination leakage current, current gain an d output conductance. Similarly, the AlN/GaN HEMTs irradiated with a dose of 2 10 11 cm 2 showed only small changes in dc transfer characteristics, threshold voltage shift, and gate lag with a high frequency pulse on the gate. Both of InAlAs/InGaAsSb/InG aAs DHBTs and AlN/GaN HEMTs appeared to be well suited for space or nuclear industry applications. 8 .5 Critical Voltage on Pt gated AlGaN/GaN Electron Mobility Transistors Significant improvement of nitride HEMT stability by using Pt based gate metallizat ion to replace the conventional Ni/Au contact was demonstrated. The off state critical voltage was increased from 55 V to >100 V, which was the instrumental limitation in this experiment. Both Schottky forward and reverse gate characteristics of the Ni/A u degraded once the gate voltage passed the critical voltage of 55 V. Minimal changes of gate and drain I V characteristics were observed for the HEMTs with Pt gate metallization. The thermal and chemical stability of the Pt contacts attribute the impro vements of the device reliability. 8 .6 Pt Gated AlGaN/GaN High Electron Mobility Transistor Based Hydrogen Sensors The effects of relative humidity on sensing characteristics of Pt gated AlGaN/GaN HEMT diode based hydrogen sensors were investigated. The absorbed water and oxygen molecules blocked available Pt surface adsorption sites for H 2 absorption and reduced the hydrogen sensing sensitivity compared to low humidity conditions. The hydrogen sensing sensitivity decreased proportional to the relative h umidity. However, the presence of humidity improved the sensor recovery characteristics after exposure to the hydrogen ambient.

PAGE 131

131 LIST OF REFERENCES 1. K. Bardeen and W. H. Brattain, Phys. Rev. 71 230 (1948). 2. W. Shockley, Bell System Technical Journal 28 4 25 (1949). 3. H. Kroemer, proc. IRE 45 1535 (1957). 4. H. T. Yuan, W. V. McLevige, and H. D. Shih, VLSI Electronics: Microstructure Science, GaAs Microelectronics, edited by W. R. Wisseman (Academic, New York, 1984). 5. D. Ankri, A. Scavennec, C. Courbet, F. He liot, and J. Riou, Appl. Phys. Lett. 40 816 (1982). 6. R. Nottenburg, J. C. Bischoff, M. B. Panish, and H. Temkin IEEE Electron Device Lett. 8 282 (1987). 7. R. N. Nottenburg, H. Temkin, M. B. Panish, R. Bhat, and J. C. Bischoff, IEEE Electron Device Lett E DL 7 643 (1986). 8. P. M. Asbeck, M. F. Chang, K. C. Wang, D. L. Miller, G. J. Sullivan, N. H. Sheng, E. Sovero, and J. A. Higgins, IEEE Transactions on Microwave Theory and Techniques MTT 35 1462 (1987). 9. Y. Oda, H. Yokoyama, K. Kurishima, T. Kobayashi, N. Watanabe and M. Uchida Appl. Phys. Lett. 87 023503 (2005). 10. C. R. Bolognesi, N. Matine, M. W. Dvorak, X. G. Xu, J. Hu, and S. P. Watkins, IEEE Electron Device Lett 20 155 (1999). 11. Walid Hafez and Milton Feng, Applied Physics Letters 86 152101 (2005). 12. H G. Liu, O. Ostinelli, Y. Zeng, and C. R. Bolognesi, Proc. IEEE International Electron Devices Meeting (IEDM 2007) 667 (2007). 13. W. Snodgrass, B. R. Wu, K. Y. Cheng, and M. Feng, Proc. IEEE International Electron Devices Meeting (IEDM 2007) 663 (2007). 14. S H Chen S Y Wang, R. J. Hsieh, and J I Chyi, IEEE Electron Device Letters 28 679 (2007) 15. S H Chen K H Teng H Y Chen S Y Wang, and J I Chyi IEEE Electron Device Lett. 29 655 (2008) 16. B. R. Wu, W. Snodgrass, M. Feng, and K. Y. Cheng, J. Crystal Growth 301 302 1005 (2007). 17. S H Chen S Y Wang, K H Teng and J I Chyi, Proc. IPRM 2007 141 (2007)

PAGE 132

132 18. M. Ida, K. Kurishiina, N. Watanabe, and T. Enoki, Proc. IEDM 2001 776 (2001). 19. Minoru Ida, Kenji Kurishima, KiyoShi Ishii, and Noriyuki Wa tanabe, Technical Digest of 2003 IEEE GaAs IC Symposium 211 (2003). 20. Shoji Yamahata, Kenji Kurishima, Hiroshi Its, and Yutaka Matsuoka, Technical Digest of 1995 IEEE GaAs IC Symposium 163 (1995). 21. Walid Hafez, William Snodgrass, and Milton Feng, Applied Ph ysics Letters 87 252109 (2005). 22. William Snodgrass, Walid Hafez, Nathan Harff, and Milton Feng, Proc. IEDM 2006 1 (2006). 23. Zach Griffith, Erik Lind and Mark J.W. Rodwell, Proc. IPRM 2007 403 (2007). 24. W. Hafez, Jie Wei Lai and M. Feng, Electronics Letters 3 9 1475 (2003). 25. Daekyu Yu, Kwangsik Choi, Kyungho Lee, Bumman Kim, H. Zhu, K. Vargason, J.M. Kuo, P. Pinsukanjana, and Y.C. Kao, Proc. IEDM 2004 557 (2004). 26. William Snodgrass and Milton Feng, Proc. GaAs MANTECH Conference 2008 277 (2008). 27. Zach Griffith a nd Mark J.W. Rodwell, Proc. IPRM 2006 96 (2006). 28. D. W. Scott, P. C. Chang, D. Sawdai, L. Dang, J. Wang, M. Barsky, W. Phan, B. Chan, B. Oyama, A. Gutierrez Aitken, and A. Oki, Proc. IPRM 2006 100 (2006). 29. J. S. Rieh, B. Jagannathan, H. Chen, K. T. Schone nberg, D. Angell, A. Chinthakindi, J. Florkey, F. Golan, D. Greenberg, S. J. Jeng, M. Khater, F. Pagette, C. Schnabel, P. Smith, A. Stricker, K. Vaed, R. Volant, D. Ahlgren, G. Freeman, K. Stein, and S. Subbanna, Proc. IEDM 2002 771 (2002). 30. C. R. Bologne si, H. G. Liu, N. G. Tao, L. Zheng, X. Zhang, and S. P. Watkins, Proc. IEDM 2005 779 (2005). 31. Norihide Kashio, Kenji Kurishima, Yoshino K. Fukai, Minoru Ida, and Shoji Yamahata, Japanese Journal of Applied Physics 49 04DF02 (2010) 32. W. Snodgrass, W. Hafez N. Harff, and M. Feng, Proc. IEEE Indium Phosphide and Related Materials (IPRM 2006), 1 (2006). 33. F. Schwierz, and J. J. Liou, Solid State Electronics 51 1079 (2007). 34. Z. Griffith, E. Lind, M. J. W. Rodwell, X. M. Fangt, D. Loubychevt, Y. Wut, J. M. Faste naut, and A. W. K. Liut, Proc. IEEE Indium Phosphide and Related Materials (IPRM 2007), 403 (2007)

PAGE 133

133 35. S. M. Sze, Physics of Semiconductor Devices, 2nd ed. (Wiley New York 1981) 36. D. A. Neamen, Semiconductor physics and devices, 3 rd ed. (McGraw Hill, San Fr ansisco, 2003 ). 37. H. Fukano, H. Nakajima, T. Ishibashi, Y. Takanashi, M. Fujimoto, IEEE Trans. Electron Devices 39 500 (1992). 38. M. Lundstrom, Fundamentals of Carrier Transport 326 327 (Cambridge University Press, Cambridge, U.K., 2000). 39. 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 Letters 25 117 (2004). 40. Haifeng Sun, Andreas R. Alt, Diego Marti, Mathias Vetter, Hansruedi Benedickter, and C. R. Bolognesi Applied Physi cs Express 2 111002 (2009). 41. M. Higashiwaki, T. Mimura, and T. Matsui, IEEE Electron Device Lett. 27 719 ( 2006 ). 42. E. Alekseev, A. Eisenbach, and D. Pavlidis, Electron. Lett. 35 2145 (1999). 43. H. Kawai, M. Nakamura, and S. Imanaga, Electron. Lett. 34 592 ( 1 998 ). 44. S. Imanaga and H. Kawai, Jpn. J. Appl. Phys., Part 1 39 1678 (2000). 45. A. M. Dabiran, A. M. Wowchak, A. Osinsky, J. Xie, B. Hertog, B. Cui, D. C. Look, and P. P. Chow, Appl. Phys. Lett. 93 082111 (2008). 46. Haifeng Sun Andreas R. Alt Hansruedi Benedic kter, Eric Feltin Jean Franois Carlin, Marcus Gonschorek, Nicolas Grandjean, and C. R. Bolognesi IEEE Electron Device Letters 31 293 (2010). 47. Intel Corporation, website: www.intel.com 48. M N. Yoder, IEEE Trans. on Ele ctron Dev. 43 1633 (1996) 49. S. J. Pearton, F. Ren, A. P. Zhang, and K. P. Lee, Mat. Sci. and Eng. R 30 55 (2000). 50. Safa O. Kasap, and Peter Capper, Springer handbook of electronic and photonic materials 54 (Springer, New York, 2006). 51. R.E. Williams, Modem GaAs Processing Methods (Artech House, Boston, 1990). 52. J.R. Lothian, F. Ren, S.J. Pearton, U.K. Chakrabarti, C.R. Abernathy, and A. Katz, J. Vac. Sci. Technol. B 10 2361 (1992). 53. A. Lebib, Y. Chen, F. Carcenac, E. Cambril, L. Manin, L. Couraud, and H. Launo is, Microelectronic Engineering 53 175 (2000).

PAGE 134

134 54. P. Rai Choudhury, Handbook of Microlithography, Micromachining, and Microfabrication SPIE Optical Engineering Press (London, UK, 1997). 55. A. Lebib, Y. Chen, E. Cambril, P. Youinou, V. Studer, M. Natali, A. Pp in, H.M. Janssen, R.P. Sijbesma, Microelectronic Engineering 61 62 371 (2002). 56. H. Sun et al. IEEE Electron Device Lett. 30 796 (2009). 57. G. H. Jessen et al. IEEE Electron Device Lett 28 354 (2007). 58. M. Neuburger et al. Int. J. High Speed Electron. Syst 14 785 (2004). 59. P. Kordos et al. IEEE Electron Device Lett 31 180 (2010). 60. 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). 61. K. D. Chabak et al. IEE E Electron Device Lett 31 561 (2010). 62. 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). 63. F. Medjdoub, M. Alomari, J. F. Carlin, M. Gonschorek, E. Feltin, M. A. Py, N. Gran djean, and E. Kohn IEEE Electron Device Lett 29 422 (2008). 64. 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. 2007 2145 (2007). 65. F. Medjdoub et al. Electron. Le tt 42 779 (2006). 66. M. Higashiwaki, T. Mimura, and T. Matsui, Jpn. J. Appl. Phys., Part 2 45 L843 (2006). 67. M. Wu et al. J. Vac. Sci. Technol B 28 908 (2010). 68. J. Kuzmik et al. J. Appl. Phys 106 124503 (2009). 69. J. Joh and J. del Alamo, Proceedings of th e International Electron Devices Meeting Technical Digest 2006 45 (2006). 70. F. A. Padovani and R. Stratton, Solid State Electron 9 695 (1966). 71. D. Selvanathan, L. Zhou, V. Kumar, I. Adesida, and N. Finnegan, J. Electron. Mater. 32 335 (2003). 72. L. Voss, R. Khanna, S. J. Pearton, F. Ren, and I. I. Kravchenko, Appl. Phys. Lett. 88 012104 (2006). 73. I. Ahmad, V. Kasisomayajula, M. Holtz, J. M. Berg, S. R. Kurtz, C. P. Tigges, A. A. Allerman, and A. G. Baca, Appl. Phys. Lett 86 173503 (2005).

PAGE 135

135 74. T. Makimoto, Y. Yam auchi, and K. Kumakura, Appl. Phys. Lett 84 1964 (2004). 75. A. P. Zhang et al. Solid State Electron 47 821 (2003). 76. W. Saito, T. Domon, I. Omura, M. Kuraguchi, Y. Takada, K. Tsuda, and M. Yamaguchi, IEEE Electron Device Lett 27 326 (2006). 77. R. Wang, Y. C ai, W. Tang, K. M. Lau, and K. J. Chen, IEEE Electron Device Lett 27 633 (2006). 78. N. Tipirneni, A. Koudymov, V. Adivarahan, J. Yang, G. Simin, and M. A. Khan, IEEE Electron Device Lett 27 716 (2006). 79. Y. Dora, A. Chakraborty, L. McCarthy, S. Keller, S. P DenBaars, and U. K. Mishra, IEEE Electron Device Lett 27 713 (2006). 80. 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). 81. D. M. Fanning et al. Ga As MANTECH Technical Digest 2005 (2005). 82. A. Minko et al. IEEE Electron Device Lett 25 453 (2004). 83. M. Higashiwaki, T. Mimura, and T. Matsui, IEEE Electron Device Lett 27 719 (2006). 84. E. Alekseev, A. Eisenbach, and D. Pavlidis, Electron. Lett 35 2145 ( 1999). 85. H. Kawai, M. Nakamura, and S. Imanaga, Electron. Lett 34 592 (1998). 86. S. Imanaga and H. Kawai, Jpn. J. Appl. Phys., Part 1 39 1678 (2000). 87. A. M. Dabiran, A. M. Wowchak, A. Osinsky, J. Xie, B. Hertog, B. Cui, D. C. Look, and P. P. Chow, Appl. Phys. Lett 93 082111 (2008). 88. N. Moser et al. J. Electrochem. Soc. 151 G915 (2004). 89. B. P. Gila et al. Solid State Electron 50 1016 (2006). 90. K. N. Lee, S. M. Donovan, B. Gila, M. Overberg, J. D. MacKenzie, C. R. Abernathy, and R. G. Wilson, J. Electrochem. Soc 147 3087 (2000). 91. C. Y. Chang, S. J. Pearton, C. F. Lo, F. Ren, I. I. Kravchenko, A. M. Dabiran, A. M. Wowchak, B. Cui, and P. P. Chow Appl. Phys. Lett 94 263505 (2009). 92. N. X. Nguyen, M. Micovic, W. S. Wong, P. Hashimoto, L. M. McCray, P. Janke, C. Nguyen, Electron. Lett 36 468 (2000). 93. S. J. Pearton, J. C. Zolper. R. J. Shul and F. Ren, J. Appl. Phys ., 86 1 (1999).

PAGE 136

136 94. 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). 95. I. Daumiller, C. Kirchner, M. Kamp, K. J. Ebeling, and E. Kohn, IEEE Electron Device Lett 20 448 (1999). 96. B. M. Green, K. K. Chu, E. M. Chumbes, J. A. Smart, J. R. Shealy, and L. F. Eastman, IEEE Electro n Device Lett 21 268 (2000). 97. G. Simin, A. Koudymov, A. Tarakji, X. Hu, J. Yang, M. Asif Khan, M. S. Shur, and R. Gaska, Appl. Phys. Lett 79 2651 (2001). 98. B. Luo, R. Mehandru, J. Kim, F. Ren, B. P. Gila, A. H. Onstine, C. R. Abernathy, S. J. Pearton, R. Fitch, J. Gillespie, T. Jenkins, J. Sewell, D. Via, A. Crespo and Y. Irokawa, J. Electrochem. Soc 149 G613 (2002). 99. E. M. Chumbes, J. A. Smart, T. Prunty, and J. M. Shealy, IEEE Trans. Electron Devices 48 416 (2001). 100. B. Luo, J. W. Johnson, J. Kim, R. M. Mehandru, F. Ren, B. P. Gila, A. H. Onstine, C. R. Abernathy, S. J. Pearton, A. G. Baca, R. D. Briggs, R. J. Shul, C. Monier, and J. Han, Appl. Phys. Lett 80 1661 (2002). 101. P. G. Neudeck and C. Fazi, IEEE Trans. Electro n Devices 46 485 (1999). 102. A. P. Zhang, L. B. Rowland, E. B. Kamensky, V. Tilak, J. C. Grande, J. Feetsov, A. Vatiakhikh, and L. F. Eastman, J. Electron. Mater 32 626 (2003). 103. C. F. Lo., T.S. Kang, C.Y. Chang, S.J. Peaton, I.I. Kravchenko, O. Laboutin, J. W. Johnson, and F. Ren, Appl. Phys. Lett 97 262116 (2010). 104. T. Kimoto, N. Miyamoto, and H. Matsunami, IEEE Trans. Electron Devices 46 471 (1999). 105. P. G. Neudeck and C. Fazi, IEEE Trans. Electron Devices 46 485 (1999). 106. F. Ren and J.C. Zolper, Wide Energy Bandgap Electronic Devices (World Scientific Pub. Co. Inc., 2003). 107. L. Do Thanh, V. Exner and P. Balk, Appl. Sur. Sci 30 204 (1987). 108. B. Luo et al. J. Electrochem. Soc 149 G213 (2002). 109. X. Hu et al. IEEE Trans. Nucl. Sci 49 3213 (2002). 110. S. Zhang, G. N iu, J. D. Cressler, S. J. Mathew, U. Gogineni, S. D. Clark, P. Zampardi, and R. L. Pierson, IEEE Trans. Nucl. Sci 47 2521 (2000).

PAGE 137

137 111. Y. Song, M. E. Kim, A. K. Oki, and M. E. Hafizi, IEEE Trans. Nucl. Sci 36 2155 (1989). 112. S. Vuppala, C. Li, P. Zwicknagl, an d S. Subramanian, IEEE Trans. Nucl. Sci 50 1846 (2003). 113. H. Y. Kim, Jihyun Kim, S. P. Yun, K. R. Kim, T. J. Anderson, F. Ren, and S. J. Pearton, J. Electrochem. Soc 155 H513 (2008). 114. J. J. Suter, J. M. Cloeren, J. R. Norton, and D. Y. Kusnierkiewicz, IEE E Trans. Nucl. Sci 34 1070 (1987). 115. S. J. Cai et al. IEEE Trans. Electron Devices 47 304 (2000). 116. B. Luo et al. J. Electron. Mater 31 437 92002). 117. B. Luo et al. Appl. Phys. Lett 79 2196 (2001). 118. R. Neuberger, G. Mller, O. Ambacher, and M. Stutzmann, Phys. Status Solidi A 183 R10 (2001). 119. B. Luo et al. Appl. Phys. Lett 80 604 (2002). 120. C. Claeys and E. Simoen, Radiation Effects in Advanced Semiconductor Materials and Devices 145 (Springer Verlag, Berlin, 2002). 121. B. Luo et al. Solid State Electron 4 7 1015 (2003). 122. X. Hu et al. IEEE Trans. Nucl. Sci 50 1791 (2003). 123. 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). 124. X. Hu et al. IEEE Tr ans. Nucl. Sci. 51 293 (2004). 125. H. Y. Kim, J. Ahn, J. Kim, S. P. Yun, and J. S. Lee, J. Ceram. Proc. Res 9 155 (2008). 126. K. N. Lee, S. M. Donovan, B. Gila, M. Overberg, J. D. MacKenzie, C. R. Abernathy, and R. G. Wilson, J. Electrochem. Soc 147 3087 (200 0). 127. A. Ionascut Nedelcescu, C. Carlone, A. Houdayer, H. J. von Bardeleben, J. L. Cantin, and S. Raymond, IEEE Trans. Nucl. Sci 49 2733 (2002). 128. Y. F. Wu, A. Saxler, M. Moore, R. P. Smith, S. Sheppard, P. M. Chavarkar, T. Wisleder, U. K. Mishra, and P. Par ikh, IEEE Electron Device Lett 25 117 (2004). 129. H. F. Sun, A. R. Alt, H. Benedickter and C. R. Bolognesi, Electronics Letters 45 376 (2009).

PAGE 138

138 130. N. X. Nguyen, M. Micovic, W. S. Wong, P. Hashimoto, L. M. McCray, P. Janke and C. Nguyen, Electron. Lett 36 468 (2000). 131. S. J. Pearton, J. C. Zolper. R. J. Shul and F. Ren, J. Appl. Phys 86 1 (1999). 132. 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). 133. H. Kim, V. Tilak, B. M. Green, H. Cha, J. A. Smart, J. R. Shealy and L. F. Eastman, Proceedings of International Reliability Physics Symposium 2001 214 (2001). 134. 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). 135. D. Pavlidis, P. Valizadeh and S. H. Hsu, Proceedings of GaAs Symposium 2005 265 (2005). 136. J. Joh and J. A. del Alamo, IEEE IEDM Technol Digest 2006 1 (2006). 137. J. Joh, L. Xia, J. A. del Alamo, IEEE IEDM Tech nol Digest 2007 385 (2007). 138. G. Meneghesso, G. Verzellesi, F. Danesin, F. Rampazzo, F. Zanon, A. Tazzoli, M. Meneghini, E. Zanoni, IEEE Trans. on Dev. and Mat. Reliability 8 332 (2008). 139. J. Joh and J. A. del Alamo, IEEE Electron Dev. Lett 29 287 (2008). 140. J. A. del Alamo and J. Joh, Microelectronics Reliability 49 1200 (2009). 141. S. Demirtas, J. Joh and J. A. del Alamo, Microelectronics Reliability 50 758 (2010). 142. E. F. Chor, R. J Malik, R. A. Hamm, and R. Ryan, IEEE Electron Dev. Lett 17 62 (1996). 143. H. Ok ada, S. I. Shikata, and H. Hayshi, Jap. J. Appl. Phys 30 L558 (1990). 144. J. R. Lothian, F. Ren, J. M. Kuo, J. S. Weiner and Y. K. Chen, Solid State Ekctron ics 41 673 (1997). 145. M. Dammann, A. Leuther, F. Benkhelifa, T. Feltgen, and W. Jantz, P hys. Stat. Sol. (a) 195 81 (2003). 146. K. J. Chen, T. Enoki, K. Maezawa, K. Arai, and M. Yamamoto, IEEE Trans. On Electron Dev 43 252 (1996). 147. E. Monroy, F. Calle, R. Ranchal, T. Palacios, M. Verd, F. J. Snchez, M. T. Montojo, M. Eickhoff, F. Omns, Z. Bougrioua and I. Mo erman, Semiconductor Science and Technology 17 L47 (2002).

PAGE 139

139 148. J. Pedrs, 1, R. Cuerdo1, R. Lossy, N. Chaturvedi, J. Wrfl, and F. Calle, Phys. Stat. Sol. (c) 3 1709 (2006). 149. S. Karmalkar, N. Satyan, and D. M. Sathaiya IEEE Electron Dev. Lett 27 87 (2006). 150. J. Kuzmik, D. Pogany, E. Gornik, P. Javorka and P. Kordos, Solid State Electronics 48 271 (2004). 151. Appl. Ohys. Lett 97 023503 (2010). 152. Engin Arslan, Serkan Btn, and Ekmel Ozbay, Applied Physics Letters 94 1421 06 (2009). 153. Chih Yang Chang, E.A. Douglas, Jinhyung Kim, Liu Lu, Chien Fong Lo, Byung Hwan Chu, D.J. Cheney, B.P.Gila, F.Ren, G.D.Via, David A. Cullen, Lin Zhou, David. J. Smith, Soohwan.Jang and S.J.Pearton, IEEE Trans. Electron & Mat. Reliab 11 187 (201 1). 154. M. Eickhoff, J. Schalwig, G. Steinhoff, O. Weidmann, L. Gorgens, R. Neuberger, M. Hermann, B. Baur, G. Muller, O. Ambacher, and M. Stutzmann, P hysica Status Solidi C 6 1908 (2003). 155. A. Baranzahi, A. Lloyd Spetz, and I. Lundstrm, Appl. Phys. Lett 67 3203 (1995). 156. J. Schalwig, G. Muller, O. Ambacher, and M. Stutzmann, Phys. Status Solidi A 185 39 (2001). 157. J. Kim, F. Ren, B. Gila, C. R. Abernathy, and S. J. Pearton, Appl. Phys. Lett 82 739 (2003). 158. J. Kim, B. Gila, G. Y. Chung, C. R. Abernathy, S. J. Pe arton, and F. Ren, Solid State Electron. 47 1069 (2003). 159. J. R. Huang, W. C. Hsu, Y. J. Chen, T. B. Wang, K. W. Lin, H. I. Chen, and W. C. Liu, Sens. Actuators B 117 151 (2006). 160. B. S. Kang, S. Kim, F. Ren, B. P. Gila, C. R. Abernathy, and S. J. Pearton, S ens. Actuators B 104 232 (2005). 161. K. Matsuo, N. Negoro, J. Kotani, T. Hashizume, and H. Hasegawa, Appl. Surf. Sci 244 273 (2005). 162. http://www.cityrating.com/relativehumidity.asp 163. L. Voss, B. P. Gila, S. J. Pearton, H. T. Wang, and F. Ren, J. Vac. Sci. Technol. B 23 2373 (2005).

PAGE 140

140 164. H. T. Wang, B. S. Kang, F. Ren, R. C. Fitch, J. Gillespie, N. Moser, G. Jessen, R. Dettmer, B. P. Gila, C. R. Abernathy, and S. J. Pearton, Appl. Phys. Lett 87 17210 5 (2005). 165. T. J. Anderson, H. T. Wang, B. S. Kang, F. Ren, S. J. Pearton, A. Osinsky, Amir Dabiran, and P. P. Chow, Applied Surface Science 255 2524 (2008). 166. J. Huang, W. C. Hsu Y. J Chen, T. B. Wang, K. W. Lin, H. I. Chen, and W. C. Liu, Sensors and Actu ators B117 151 (2006). 167. S. Y. Chiu, H. W. Huang, K. C. Liang, T. H. Huang, K. P. Liu, J. H. Tsai and W. S. Lour, Semicond. Sci Technol 24 045007 (2009). 168. B. E. Conway and B. Barnett, H. Angerstein Kozlowska, and B. V. Tilak, J. Chem. Phys 93 8361 (199 0). 169. D. A. Harrington, J. Electroanal. Chem 420 101 (1997). 170. P. N. Ross, Surf. Sci. Rep 45 121 (2002). 171. Electrochemical and Solid State Letters 7 A82 (2004). 172. H. Xu, R. Kunz, and J. M. Fenton, Electrochemical and Solid State Letters 10 B1 (2007). 173. H. T. Wang, T. J. Anderson, B. S. Kang, F. Ren, Changzhi Li, Z. N. Low, J. Lin, B. P. Gila, S. J. Pearton, A. Osinsky, and Amir Dabiran, Appl. Phys. Lett 90 252109 (2007). 174. H. T. Wang, T. J. Anderson F. Ren, C. Z. Li, Z. N. Low, J. Lin, B. P. Gila and S. J. Pearton Appl. Phys. Lett 89 24211 (2006). 175. F. C. Huang, Y. Y. Chen and T. T. Wu, Nanotechnol 20, 065501 (2009). 176. Z. Zhao, M. Knight, S. Kumar, E. T. Eisenbraun, and M. A. Carpenter, Sensors an d Actuators B129 726 (2008). 177. S H Chen C. M. Chang, P. Y. Chiang, S Y Wang, W. H. Chang, and J I Chyi, IEEE Transactions on Electron Devices 57 3327 (2010).

PAGE 141

141 BIOGRAPHICAL SKETCH Chien Fong Lo was born in ChangHua Taiwan, in 197 8 He received both the b achelor and m aster degree in the Department of Chemical Engineering at the National Taiwan University, Taipei Taiwan, in 2001 and 200 3, respectively. He attended the Chemical Warfare Corps of Taiwan Army for the mandatory training and service, and severed as vice company commander with the position of 3rd lieutenant officer from 200 4 to 200 6 After military service, he worked as a research assistant in the Department of Materials Science and Engineering at the National Taiwan University until 2 007 In 200 7 f all, he enrolled in the Master program in the Department o f Chemical Engineering at the University of Florida, and enrolled in the Ph.D. program in the same d epartment at the University of Florida in 2008 fall He joined Professor research group and began the substantive research used to develop this dissertation. U nder the guidance of Prof Fan Ren he focused on the wide bandgap semiconductor materials and sensors, high speed InP based heterojunction bipolar transistors GaN base d high electron mobility transistors, as well as electronic device reliability I n May 2011 h e earned his D octor of P hilosophy degree from the Chemical Engineering Department of the U niversity of Florida with more than 26 SCI journal publications in high ly recognized journals ( Applied Physics Letters Electrochemical and Solid State Letters Sensors and Actuators IEEE Transactions on Device and Materials Reliability J ournal of Vacuum Science & Technology etc. ), and 9 international conference presentati ons.

PAGE 142

142 FABRICATION AND CHARACTERIZATION OF HETEROJUNCTION TRANSISTORS Candidate's name: Chien Fong Lo Phone number and e mail address : 352 871 3995 / cflo@ufl.edu Department: Chemical Engineering Supervisory committee chair: Fan Ren Degree: Doctor of Philo sophy Month and year of graduation: May 2011 The main objective of this work were focus ed on the fabrication and characteriz ation for both of InAlAs/InGaAsSb/InGaAs double heterojunction bipolar transistors (DHBTs) and GaN based high electron mobility tr ansistors (HEMTs) including AlGaN/GaN, InAlN/GaN, and AlN/GaN HEMTs. Ultra low contact resistance after a short term annealing on nov e l InGaAsSb base of DHBTs was achieved and the excellent dc and rf performance were demonstrated. E ffects of the Ohmic contact annealing for lattice matched InAlN/GaN HEMTs with and without a thin GaN cap layer were exhibited and their optimal annealing temperature were obtained Oxide passivation on AlN/GaN HEMTs with the UV ozone treatment and SiN x passivation on AlGaN/ GaN HEMTs with the PECVD were investigated. Furthermore, proton radiation induced degradation on InAlAs/InGaAsSb/InGaAs DHBTs and AlN/GaN HEMTs was demonstrated, while t he radiation hardness of HBTs and HEMTs is one of the critical factors that need to be established for military, space, and nuclear industry applications For the off state step stress on AlGaN/GaN HEMTs with Ni/Au and Pt/Ti/Au gate metallization a significant improvement of critical voltage from 55 V to over 100 V on Pt gated devices w as observed. Finally, the effects of ambient on the Pt gated AlGaN/GaN HEMT sensor for hydrogen gas sensing application were also explored in this dissertation.