CHARACTERIZATION AND PROCESS DE VELOPMENT OF ZINC OXIDE-BASED LIGHT-EMITTING DIODES By JAU-JIUN CHEN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2006
Copyright 2006 by Jau-Jiun Chen
Dedicated to my family and all the pe ople who have supported me during my life
iv ACKNOWLEDGMENTS I would first like to acknowledge my academic advisor, Dr. Fan Ren, for his invaluable guidance in the field of semiconduc tors. He has always been very generous with his time and expertise. He has enlight ened my life as my mentor and helped me grow professionally. I would al so like to thank Dr. Jason F. Weaver for leading me into the field of surface science. I would also li ke to extend my gratitude to Dr. Stephen J. Pearton and Dr. David P. Norton. Their support and delightful conversations bore fruit in my work. There are many colleagues to thank for their assistance with experiments including Dr. Chih-Yang Chang, Dr. Chi-Joe Kao, Khanna Rohit, Dr. Byoung Sam Kang, Soohwan Jang, Travis Anderson, Hung-Ta Wang, Dr. Ke lly Ip, Voss Lars, Dr. Sang-Youn Han, Dr. Luc Stanford, Jon Wright, and Wantae Lim. Gratefully, I would lik e to acknowledge Dr. Brent P. Gila for his support with scientif ic research and kind ness. I also would especially thank Dr. Yuanjie Li, Mark Hl ad, Andrew Gerger, and Hyun-Sik Kim, who grew the samples used in this study. I al so thank Santiago Taveres for the computer technical support. I thank Shirley Kelley, Nanc y, and Debbie Aldrich for their invaluable assistance and delightful conversations. I am also indebted to Eric Lamber, James Hinant, and Dennis Vince for the technical support. It has been a privilege to work with all of them. I also thank the friends who have supported me in my graduate student life. I am grateful to thank Hui-Yi Li for her inspira tion and kindness, as my best friend. I thank
v Dr. Yangling Chou for her friendship and assi stance in accommodati ng to life here in Gainesville. I extend my gratitude towa rd Dr. Ben-Zu Wan for his thoughts and encouragement. Special recognition is dese rved for my language instructors, Enid Corbin and Yuko Uchida, for their assistance in Gainesville and fo r bringing color into my life. I especially tha nk Dr. Kehuey Wu and my brother, Jau-Nan Chen, for their fruitful discussions in the fi eld of electrical engineering. My conversations with them have been most delightful and will be remembere d. It is really great to indulge in science like this. I would like to al so thank Jia-Wha Wang, Pao-Yun Tao, Jane Phipps, Dr. Anne Heaphe, Chia-Bin Feng, Makiko Kasahara, Mei-Hung Chu, Dr. Zhi Chen, Dr. Wen-Ben Lou, Wei Liu, Jau-Jr Lin, Dr. James Keesli ng and Tomas Marson with all their company and spiritual support. My soul mate, Dr. Alex Gerrard, is the one to whom I am most grateful. He has encouraged me, shared knowledge and discussed science with me. I truly appreciate his company and encouragement through the good and challenging times. Finally, I would like to show my gratitude to my mom a nd dad. They always have shown their unconditional love to me. I cannot e xpress my deep thoughts of gratitude.
vi TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES.............................................................................................................ix LIST OF FIGURES.............................................................................................................x ABSTRACT.....................................................................................................................xi v CHAPTER 1 INTRODUCTION........................................................................................................1 2 BACKGROUND..........................................................................................................6 2.1 ZnO Properties........................................................................................................6 2.2 Wet Etching............................................................................................................7 2.3 Metal-Semiconductor Contacts..............................................................................9 2.3.1 Rectifying Contacts....................................................................................10 2.3.2 Ohmic Contacts..........................................................................................11 2.3.3 Transmission Line Method (TLM).............................................................13 2.4 Bandgap Engineering............................................................................................15 2.5 Processing Techniques..........................................................................................17 2.5.1 E-Beam Evaporator....................................................................................17 2.5.2 Electrical Measurements............................................................................17 2.5.3 Rapid Thermal Annealing..........................................................................17 2.5.4 Sputter Deposition......................................................................................18 2.6 Characterization Techniques................................................................................18 2.6.1 Auger Electron Spectroscopy (AES)..........................................................18 2.6.2 Atomic Force Microscopy (AFM)..............................................................19 2.6.3 Photoluminescence (PL).............................................................................20 2.6.4 Profilometry................................................................................................21 2.6.5 Scanning Electron Microscopy (SEM).......................................................21 2.6.6 X-Ray Photoelectron Spectroscopy (XPS).................................................22 3 WET ETCH OF ZINC OX IDE-BASED MATERIALS............................................35 3.1 Introduction...........................................................................................................35 3.2 Experimental.........................................................................................................37
vii 3.2.1 Growth of ZnCdO.......................................................................................37 3.2.2 Growth of ZnMgO......................................................................................38 3.2.3 Wet Etch Method........................................................................................38 3.3 Results and Discussion.........................................................................................38 3.4 Summary...............................................................................................................41 4 OHMIC CONTACTS TO ZINC OXIDE-BASED MATERIALS............................52 4.1 Introduction...........................................................................................................52 4.2 Experimental.........................................................................................................55 4.2.1 Ti/Au and Ti/Al/Pt/Au Ohmic Contacts to N-type ZnCdO........................55 4.2.2 Ti/Au to Aluminum Doped ZnO................................................................57 4.2.3 ITO/Ti/Au Ohmic Contacts to Aluminum Doped ZnO.............................58 4.3 Results and Discussion.........................................................................................59 4.3.1 Ti/Au and Ti/Al/Pt/Au Ohmic Contacts to N-type ZnCdO........................59 4.3.2 Ti/Au to Aluminum Doped ZnO................................................................62 4.3.3 ITO/Ti/Au Ohmic Contacts to Aluminun Doped ZnO...............................64 4.4 Summary...............................................................................................................66 5 BANDGAP ENGINEERING.....................................................................................89 5.1 Introduction...........................................................................................................89 5.2 Experimental.........................................................................................................92 5.2.1 ZnCdO/ZnO Heterojunction Band Offset..................................................92 5.2.2 MgO/GaN Heterojunction Band Offset......................................................94 5.2.3 Sc2O3/GaN Heterojunction Band Offset....................................................95 5.3 Results and Discussion.........................................................................................97 5.3.1 ZnCdO/ZnO Heterojunction Band Offset..................................................97 5.3.2 MgO/GaN Heterojunction Band Offset......................................................98 5.3.3 Sc2O3/GaN Heterojunction Band Offset....................................................99 5.4 Summary.............................................................................................................100 6 ZINC OXIDE-BASED LIGHT-E MITTING DIODES SIMULATION..................112 6.1 Introduction.........................................................................................................112 6.2 Models and Parameters.......................................................................................113 6.2.1 Strain and Piezoeffect...............................................................................115 6.2.2 Carrier Concentration...............................................................................116 6.2.3 Radiative and Non-radiative Recombination...........................................117 6.2.4 Light Emission Efficiency........................................................................118 6.3 Results and Discussion.......................................................................................119 6.4 Summary.............................................................................................................120 7 CONCLUSIONS AND FUTURE WORK...............................................................127 APPENDIX A PYSICAL PROPERTIES OF METALS..................................................................132
viii B PHYSICAL PROPERTIES OF ZINC OXIDE........................................................133 LIST OF REFERENCES.................................................................................................134 BIOGRAPHICAL SKETCH...........................................................................................149
ix LIST OF TABLES Table page 2-1 Physical properties of ZnO.......................................................................................23 2-2 Bandgap and electron affinity of GaN and ZnO......................................................23 4-1 Concentration of elements detected in Ti/Au contacts on ZnCdO/ZnO/GaN (in Atom %)...................................................................................................................68 4-2 Concentration of elements de tected in Ti/Al/Pt/Au contacts on ZnCdO/ZnO/GaN (in Atom %)................................................................................68 5-1 Values of ZnCdO/ZnO band offsets determined in these experiments..................102 5-2 Values of MgO/GaN band offsets determined in these experiments.....................102 5-3 Values of Sc2O3/GaN band offsets determined in these experiments....................102 6-1 Simulated parameters conducted in ZnMgO/ZnCdO/ZnMgO. Bolded conditions in the table correspond to the refe renced values in Figure 6-2..............................121
x LIST OF FIGURES Figure page 2-1 Crystal structure of ZnO...........................................................................................24 2-2 Energy band diagram of a metal/n-t ype semiconductor contact at thermal equilibrium...............................................................................................................25 2-3 Energy band diagram of a metal/p-t ype semiconductor contact at thermal equilibrium...............................................................................................................26 2-4 Schematic diagram of a TLM pattern......................................................................26 2-5 Total resistance as a function of TLM pad spacing.................................................27 2-6 TLM without etch mesa (top); TLM with etch mesa (bottom)................................28 2-7 Top view of a circular TLM pattern.........................................................................29 2-8 Energy band lineups: (a ) straddling, (b) staggered, and (c) broken-gap..................29 2-9 Energy-band diagram at an abrupt heterojunction interface....................................30 2-10 Linear method applied to the spectrum results in the maximum valence band.......31 2-11 Operating procedure of rapid thermal annealing......................................................31 2-12 Interaction between the probe tip and substrate as a function of distance...............32 2-13 Exciton generated because the incident laser beam is greater than the bandgap.....33 2-15 XPS photoemission and Auger process...................................................................34 3-1 Etch rate of ZnCdO with diffe rent concentratio ns of HCl and H3PO4 solutions diluted in water.........................................................................................................42 3-2 Etch rate of Zn0.9Mg0.1O in different concentrations of HCl and H3PO4 diluted with water.................................................................................................................43 3-3 Arrhenius plot of ZnCdO etch rate in 0.0031M HCl and 0.0029M H3PO4 solutions diluted in water.........................................................................................44
xi 3-4 Arrhenius plot of Zn0.9Mg0.1O etch rate in 0.024M concentrations of HCl and H3PO4 diluted with water.........................................................................................45 3-5 Etch rate of ZnO in different concentrations of HCl and H3PO4 diluted with water.........................................................................................................................4 6 3-6 Arrhenius plot of ZnO etch rate in 0.24M HCl and 0.06M H3PO4 diluted with water.........................................................................................................................4 7 3-7 Etch selectivity of ZnCdO to ZnO at room temperature as a function of solution concentration............................................................................................................48 3-8 Etch selectivity of ZnMgO to ZnO as a function of solution concentration............49 3-9 Optical microscopy image (top) of ZnCd O selectively etched with HCl solution from an underlying ZnO layer. The photoresist mask is still in place. At bottom is an SEM image of the selective rem oval of ZnCdO from an underlying ZnO substrate....................................................................................................................50 3-10 SEM images of ZnMgO selectively etch ed with an underlying ZnO layer. (a) The overall pattern with a 3000 X mag. (b) a 3700 X mag image showing the ZnMgO sidewall.......................................................................................................51 4-1 Cross-sectional TEM image of as -grown Al-doped ZnO on MgO buffer on sapphire substrate.....................................................................................................69 4-2 TEM cross-section of ZnCdO layers grown on a ZnO buffer layer deposited on GaN using sapphire as a substrate. The SADPÂ’s from individual layers are also shown.......................................................................................................................70 4-3 Higher magnification TEM cross-secti on images of the as-grown structure...........71 4-4 Sheet resistance as a function of annealing temperature for Ti/Auand Ti/Al/Pt/Au contacts on ZnCdO...............................................................................72 4-5 Transfer resistance as a function of annealing temperat ure for Ti/Au and Ti/Al/Pt/Au contacts on ZnCdO...............................................................................73 4-6 Specific contact resistivity as a functi on of annealing temperature for Ti/Au and Ti/Al/Pt/Au contacts on ZnCdO...............................................................................74 4-7 Schematic of proposed contact cond uction mechanism, through reduction of depletion region by formation of oxyge n vacancies and an interfacial TiOx layer..75 4-8 Optical microscopy images of Ti/Au (top) and Ti/Al/Pt/Au contacts (bottom) annealed at differe nt temperatures...........................................................................76
xii 4-9 AES surface scans of Ti/Au contacts on ZnCdO/ZnO/GaN as-received (bottom), after annealing at 450oC (middle), and 500oC (top).................................................77 4-10 AES depth profiles on Ti/Au cont acts on ZnCdO/ZnO/GaN as-received (bottom), after annealing at 450oC (middle), and 500oC (top).................................78 4-11 AES surface scans of Ti/Al/Pt/Au contacts on ZnCdO/ZnO/GaN as-received (bottom), after annealing at 450oC (middle), and 500oC (top).................................79 4-12 AES depth profiles on Ti/Pt/Al/Au contacts on ZnCdO/ZnO/GaN as-received (bottom), after annealing at 450oC (middle), and 500oC (top).................................80 4-13 Electrical properties of Ti/Au contacts on ZnO:Al. (a) Sheet resistance as a function of annealing temperature. (b) Transfer resistance as a function of annealing temperature..............................................................................................81 4-14 Specific contact resistivity as a func tion of annealing temperature for Ti/Au contacts on ZnO:Al..................................................................................................82 4-15 Optical microscopy images of Ti/Au cont acts annealed at different temperatures on ZnO:Al (N ~ 9x1018 cm-3)...................................................................................83 4-16 Topographic images of (a) as rece ived Ti/Au deposited on ZnO:Al(N ~ 9.x1018 cm-3.) and (b) after a subsequent anneal to 250 oC. The scan size of these AFM images is 5 Âµm x 5 Âµm.............................................................................................83 4-17 AES surface scans (top) and depth pr ofiles (bottom) of Ti/Au contacts on ZnO:Al/sapphire as-received and after annealing at 250oC.....................................84 4-18 Sheet resistance (square da ta points) and specific contac t resistivity (circle data points) as a function of annealing te mperature for ITO/Ti/Au contacts on ZnO:Al.....................................................................................................................85 4-19 Optical microscopy images of ITO/T i/Au contacts annealed at different temperatures on ZnO:Al (N ~1.3x1019 cm-3.) The inner contact diameter is ~300 Âµ m...................................................................................................................86 4-20 AES surface scans of ITO/Ti/Au contacts on Al-doped ZnO as-received (bottom), after annealing at 350oC (middle), and 450oC (top).................................87 4-21 AES depth profiles on ITO/Ti/Au contacts on Al-doped n-ZnO as-received (bottom), after annealing at 350oC (middle), and 450oC (top).................................88 5-1 Core level survey spectra of Sc2O3, 3 nm layer of Sc2O3 on GaN/sapphire, and a GaN/sapphire template using a pass ener gy of 89.45 eV at take-off angle of 45o.103 5-2 XPS Zn 2p3 narrow scan and valence ba nd spectrum of 0.1 ÂµmZnCdO/0.1 ÂµmZnO/MOCVD GaN/C-plane sap phire and ZnO substrate................................104
xiii 5-3 Core level survey spectra of ZnCdO, 1nm layer of ZnCdO on ZnO, and a ZnO substrate using a pass energy of 187.85eV at take-off angle of 65o.......................105 5-4 Energy band diagram of thin Zn CdO/ZnO heterojunction interface. EB is the corresponding core level separati on measured across the interface.......................106 5-5 XPS Ga 3d narrow scan and valenc e band spectrum of 40 nm MgO/MOCVD GaN/C-plane sapphire and GaN/sapphire template...............................................107 5-6 Core level survey spectra of MgO, 5 nm layer of MgO on GaN/sapphire, and a GaN/sapphire template using a pass ener gy of 44.75 eV at take-off angle of 45o.108 5-7 Energy band diagram of thin MgO/GaN heterojunction interface. EB is the corresponding core level separati on measured across the interface.......................109 5-8 XPS Ga 3d and Sc 3p narrow scans and valence band spectra of 40 nm Sc2O3/MOCVD GaN/C-plane sapphire a nd GaN/sapphire template.....................110 5-9 Energy band diagram of thin Sc2O3/GaN heterojunction interface........................111 6-1 Epitaxial film on a substrate...................................................................................122 6-2 Schematic 2D view of ZnO-based LED structure (top) and its band diagram (bottom)..................................................................................................................123 6-3 Simulated emission spectra from ZnMgO/ZnCdO/ZnMgO structure as a function of both active layer thickness (top) and Cd composition (bottom)..........124 6-4 Experimentally observed change of room-temperature cathodoluminescence emission energy (solid circle) and simula ted bowing parameters (dashed line) as a function of Cd mole fraction...............................................................................125 6-5 Simulation of LED behavior fo r ZnMgO/ZnCdO/ZnMgO structures...................126
xiv Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy CHARACTERIZATION AND PROCESS DE VELOPMENT OF ZINC OXIDE-BASED LIGHT-EMITTING DIODES By Jau-Jiun Chen August 2006 Chair: Fan Ren Cochair: Jason F. Weaver Major Department: Chemical Engineering ZnO-based materials have great potential for UV light-emitting diodes (LEDs) and transparent electronics because of the high exciton binding energy of ZnO relative to GaN. Fabricating an effective LED from novel materials requires a detailed knowledge of the band offset, etch, and c ontact behavior of the material . This work determined the valence and conduction band offsets for Zn0.95Cd0.05O/ZnO (0.17 eV, 0.30 eV) and related materials using x-ray photoelectr on spectroscopy (XPS) and photoluminescence (PL). These methods were also used to study carrier confinement in two promising passivation materials: MgO/ GaN (1.06 eV, 3.30 eV) and Sc2O3/GaN (0.42 eV, 2.14 eV). To form an LED mesa, it is critical to understa nd the etch rate of ZnO-based materials. In this work, HCl and H3PO4 were used as etchants for ZnCdO/ZnO (~50 nmÂ·min-1 HCl/~15 nmÂ·min-1 H3PO4) and ZnMgO/ZnO (300-1100 nmÂ·min-1 HCl/120-300 nmÂ·min-1 H3PO4). A high degree of selectivity was sought using these etchants on ZnCdO/ZnO (~50 HCl/~15 H3PO4) and ZnMgO/ZnO (~300-400 HCl/~25 H3PO4). Alloyed Ti/Au and
xv Ti/Al/Pt/Au contacts were deposited on n-type Zn0.95Cd0.05O, with excellent contact resistivities of and 2.3 10-4 and 1.6 10-4 -cm2, respectively. Alloyed Ti/Au and indium-tin-oxide (ITO)/Ti/Au metallization of n-type Al-doped ZnO was used to create ohmic metal contacts with excelle nt contact resistivities of 6 10-8 and 4.6 10-6 -cm2. Using SiLENSe Software, the optimum active layer thickness was found to be 200 nm.
1 CHAPTER 1 INTRODUCTION The first light-emitting diode (LED) reported by Henry Round in 1907 was a Schottky diode, rather than a p-n-juncti on diode [1-4]. LEDs have been studied intensively for three decades. High-intens ity red, green, and yellow LEDs have been developed. By contrast, although blue LEDs have been developed, the technology is still maturing. Blue is one of the three primary colors (red, green, and blue). In order to produce the full color spectrum a nd white light, a blue LED is required. SiC, ZnSe, GaN, and ZnO are available materials for making a blue LED . GaN, ZnSe, and ZnO are direct bandgap (2.82-3.4 eV) materials, while SiC is an indirect bandgap (2.2-3.5 eV) material and has a low luminescence output (10~20 mcd) . Si C-based LEDs have a low intensity compared to direct bandgap ma terials. Due to the properties of direct energy bandgap materials, high-intensity light obtained from GaN, ZnSe, and ZnO is more suitable for optoelectronic devices. ZnSe -based materials have been grown at room temperature. The lifetimes of ZnSe material-based LEDs and laser diodes (LDs) are short. This is due to the high defect density (104 cm-2) in II-VI crystals, which leads to early device failure. The bandgap of Ga N, AlN, and InN is 3.4, 6.3, and 2.0 eV, respectively . A high-intensity GaN-base d LED from the ultraviolet to the yellow light region can be obtaine d by creating alloys with GaN, AlN, and InN. Although GaN-based LEDs have been st udied for 20 years, the GaN crystal material has a large number of de fects due to the lack of a lattice matched substrate. In addition, fabricating p-GaN is not a trivial process. In 1991, S. Nakamura et al. of Nichia
2 Co. developed a buffer layer at low temper ature using amorphous GaN, and then grew the GaN at high temperature on top of the bu ffer layer . This greatly improved the GaN crystal quality. S. Nakamura et al. also found that Mg can act as an acceptor only after activation by annealing Mg-doped GaN at 700Â°C in N2 . The end result was the generation of low-resistivity p-type GaN. In 1993, the NiChia Co. developed a 1Cd GaN LED with a lifetime of around 10,000 hours. In 1994, S. Nakamura et al. made InGaN/AlGaN he terostructure-based blue LEDs on a sapphire substrate; this de vice had an emission wavelength of 400 nm. In 1995, he made InGaN single-quantum-wellstructure LEDs [5,7]. Toyota Gosei and Matsushita/Panasonic cooperated to sell blue LEDs. Cree also makes a GaN LED, which is grown on SiC substrate. NiChia, Sanyo, Hewlett-Packar d, Temic, OSRAM Opto, and Lumileds also sell GaN-based blue LEDs . Despite the advances in GaN-based technology, other materials are being researched to generate more stable LEDs with a lower cost of production. ZnO is a promisi ng material for this application and is the subject of this dissertation. Before its use in semiconductors, ZnO was used as pigment in paint, ointment, creams, and lotions. Later, it was studied as a substrate for developing high-quality GaN. ZnO is a direct, wide bandgap material with a wu rtzite crystal structur e; it is used in gas sensors, transparent electr odes, liquid crystal displays, solar cells, piezoelectric transducers, photoelectronics material devices, blue and UV LEDs, and laser diodes . As of 2003, the use of ZnO in blue LED s and thin film transistors (TFT) has been studied extensively. Compared to GaN se miconductors, ZnO has the advantage of a relative low growth temperature on cheap glass and a much higher exciton binding
3 energy (~ 60 meV) than GaN (25 meV). This means that ZnO has a more stable exciton state at room temperature, since the heat energy is around 26 meV. The excitons in a ZnO semiconductor will not dissociate into free el ectrons or holes as a result of the heat at room temperature or the scattering betw een excitons. In a ddition, a commercial ZnO substrate is available. The ZnO system also involves simpler processing relative to GaN, which cannot be wet-etched in conventional acid mixtures at safe temperatures. The curvature of the conduction band and valence band for ZnO is smaller than for GaN. This means that the electron effective mass in ZnO is larger than in GaN. To emit light, it is vital to eliminate the non-radiative recombination of electron hole pairs. The intrinsic defects will be deeper , and the electron and hole mobili ty will be smaller than in GaN . D. C. Look et al . studied ZnO as a GaN substrate and developed a p-n junction for a LED using p-type ZnO [13,14]. Some groups have published results on Nor P-doped ZnO p-n junction LEDs. A. Tsukazaki et al . studied a ZnO p-i-n homojunction structure on (0001) ScAlMgO4 grown by laser molecular beam epitaxy (MBE) [15,16]. J.-H. Lim et al. has made a p-n homoj unction ZnO LED on sapphire using r-f sputtering. S. J. Jiao et al . demonstrated a ZnO p-n junction LED on a-plane Al2O3 substrate using plasma -assisted MBE . To improve ZnO LED device performance, the development of low-resistance ohmic contacts on ZnO is essential. C onventional contact formation involves the deposition of a metal contact on ZnO followed by annealing at elevated temperatures. This minimizes the voltage drop at the inte rface of the metal and ZnO-based material. Some groups have found low-contact resistivity for Nand P-type ZnO on the order of
4 10-7 and 10-5 -cm2, respectively [18-20]. In this st udy, the low-contact resistivity for Ntype ZnO was found to be on the order of 10-7 -cm2. When designing a three-layer heterostructur ed LED, it is essential to know all of the information about the corresponding band o ffsets. A three-layer LED consists of a light-emitting region, sandwiched between two layers with a bandgap wider than the center layer. The two out er layers are referred to as the Â“cladÂ” layers, whereas the center layer is referred to as the active layer. Th e clad layers confine electron hole pairs within the active layer, increasing th e light emission efficiency for the device. To determine these properties, the energy bandgap was dete rmined using photoluminescence (PL). Xray photoelectron spectroscopy (XPS) was us ed to determine the conduction band and valence band offset, as di scussed in Chapter 5. To design a ZnO-based LED, it is essential to optimize all of the dimensions and parameters associated with LED performa nce. ZnO-based LED structures were simulated as a function of thickness, doping, and composition using the SiLENSe simulation program. The simulation of Zn MgO/ZnCdO/ZnMgO shows that intensity increases with increasing ZnCdO thickness and Cd composition. Currently, high-quality p-type ZnO-based material has proved difficult to achieve in practice due to a strong self-compensation effect arising from native defects involving O vacancies or hydrogen impurities. To fabri cate a ZnO LED, an etch process is required to develop mesa structures. The etch result s are presented in Chapter 3. Furthermore, low-resistivity contacts are needed. Chap ter 4 discusses the development of these contacts using various metal schemes a nd by changing the doping density. When designing an LED, sound knowledge of the bandga p is needed to e ngineer electron hole
5 pair barriers and quantum wells. These results are presented in Chapter 5. To efficiently design a ZnO-based LED, a general understand ing of LED performance is needed. Chapter 6 discusses these results as predicted by simulation.
6 CHAPTER 2 BACKGROUND 2.1 ZnO Properties Zinc oxide has several useful physical propert ies that are ideal for its application in LED fabrication [21-23]. The material is ch eap to process because ZnO can be etched using wet chemical means. ZnO is ideal fo r aerospace applications since it has a high resistivity to radiation damage . Currently, ZnO is used in optoelectronic devices, solar cells, and transparent electrode s; however, ZnO has great pot ential in the manufacture of high efficiency and low cost white light LEDs. ZnO has a bandgap of 3.4 eV with a free exciton binding energy of 60 meV [21,23]. It is optica lly transparent, even at hi gh doping levels. ZnO has a wurtzite crystal structure with a lattice contact a = 3.253 Ã… and c = 5.211 Ã…, giving a c / a ratio of 1.602, as depicted in Fi gure 2-1. Zinc oxide is compos ed of alternate layers of zinc and oxygen atoms in the (0001) direction ( c -axis) of the crysta l . Its physical properties are listed in Tabl e 2-1. The exciton binding energy of ZnO (~60 meV) is higher than that of GaN (25 meV), which al lows it to produce hi gher light intensities . Commercial bulk crystalline ZnO is read ily available [18,23]. These high-quality films are generated using a chemical va por phase transport method (Eagle-Picher Technologies, Miami, OK), melt growth technique (Cermet, Atlanta, GA), or hydrothermal methods. ZnO materials can be grown under laboratory conditions using molecular beam epitaxy (MBE), pulsed laser deposition (PLD), laser MBE, metal-
7 organic chemical vapor deposition (MOCVD ), and hydride or chloride vapor phase deposition (HVPE) on multiple substrates. The bandgap of ZnO can be tuned by alloying it with Mg and Cd, which are substituted on the cation site. A quantum well ZnO LED can be constructed using ZnCdO as a clad layer and ZnMgO as a barrier . The substitution of Mg for Zn increa ses the bandgap up to 3.90 eV for Zn0.67Mg0.33O while still maintaining the wurtzite structure. The substitution of Cd in the Zn atoms decreases the bandgap to 3.0 eV for Zn0.9Cd0.1O. ZnO is an intrinsic n-type material due to the O-atom vacancies, Zn interstitials, and H incorporation within the crystal dur ing growth. Undoped ZnO has unstable longterm electrical properties due to the changi ng number of O vacancies as atmospheric oxygen interacts with the surface. The n-type conductivity and stability can be improved by doping with Al, Ga, and In. To make p-type ZnO, acceptor candidates such as N, P, and As have been successfully substituted in the oxygen sites [25,26]. N (1.46 Ã…) might be the best acceptor candidate for ZnO doping, si nce its ionic size is si milar to that of O (1.38 Ã…). Atoms with a larger radii mismatch site will limit the solid solubility for the anions. In addition to N, P (2.12Ã…) and As (2.22 Ã…) have been successfully used as acceptors despite their larger radii. At this time, p-type ZnO requires further characterization. 2.2 Wet Etching Wet chemical etching has been used for 30 years to remove material or create specific features on substrates that are typica lly defined by a patterned resist coating . The wet chemical process consists of thr ee steps. First, the etchant species migrates from the bulk solution to the surface of the s ubstrate. Second, the exposed portion of the material is removed by a reaction between the etchant and substrate. Finally, the
8 dissolved reaction products diffuse away from the substrate into th e bulk solution. In order to achieve a uniform and well-controlled et ch rate, the etching solution is agitated. This helps minimize the formation of concen tration gradients in the solution during the etching process, which would decrease the et ch rates. Depending on the nature of the reaction, the etching process may produce gases in the form of bubbles, which will substantially hinder the etchant migration to ward the surface. Agitation addresses this issue by assisting in the remova l of bubbles from the surface. Further complications may arise when etching features such as trenches with small length scales. These cause the formation of localized concentration gradients in which the local etch rates are decreased. Wet etching is typically not used for feat ures with length scales below ~ 2 Âµm, and general etching profiles using this technique are isotropic so that the reaction progresses in all directions. Of the wet etching steps just describe d, species diffusion and reaction kinetics dominate the process. Diffusion-limited etch ing, as the name implies, is limited by the diffusion of the reactant to the surface or of products from the surface. FickÂ’s law of diffusion may be used to approximate the etch ra te, but it is difficult due to the associated fluid dynamics in the bulk and localized ge ometries. Reaction-lim ited etching is not limited by species diffusion, but instead depends on the reaction rate at the surface. This is a strong function of temperature, which is su bstantially easier to c ontrol. The rate for reaction-limited etching is given by kT EaKe R/ (2-1) where K is a temperature depe ndent rate constant, Ea is the activation energy, k is the Boltzmann constant, and T is the solution temperature. Etching rates are also a function
9 of the film composition and density. In some cases, they can be dependent on the crystallographic orientation. For example, th e etch rate on a FCC[ 111] structure may be slower than that on a FCC[100 ] structure since the  or ientation is more densely packed. Commonly desired etch rates are around 10 to 100 nm per minute. The selectivity of etching is also an importa nt parameter during the process. This is defined as the ratio of the etchi ng rates of differi ng materials, S r r 1 2 (2-2) where r1 is the etching rate of the film being etched and r2 is the etching rate of the masking material or the material below the f ilm. A selectivity range of 25 Â– 50 is usually reasonable. The main advantage of using a wet etchi ng process is its hi gh selectivity. Wet etching is also more cost effective and cause s less substrate damage than other methods. However, wet etching may be difficult to c ontrol, may vary with the crystallographic orientation, and may not define features down to the length scales available using other techniques. 2.3 Metal-Semiconductor Contacts Electrical contacts are required to energize or collect electrical information from a device. Ideally, these contacts should have a low resistiv ity to enhance electron flow to and from the device. When depositing a me tal on a semiconductor, the total resistance can be summed up in two compon ents: the contact resistivity c ( -cm2) and the sheet resistance Rs ( / ) [28,31]. Section 2.3.3 gives the me thodology used to determine these values. The contact resistivity is
10 cVJ V ()0 1 (2-3) where J is the current density and V is the applied voltage. Th e sheet resistance is a large contributor to the total resistance. The re lationship between the total resistance and the sheet resistance is given by  R L Wt R L Ws (2-4) where R is the resistance due to the semiconductor, is the resistivity of the semiconductor, L is the length, W is the width, and t is the thickness of the layer. Assuming that the semiconductor film is smooth and of uniform thickness, Rs should not vary. 2.3.1 Rectifying Contacts A rectifying contact is typi cally formed as a metal and semiconductor are brought together (see Figures 2-2 and 2-3) . This will prevent most electrons and holes from flowing freely between the contact and semic onductor. The rectifyi ng behavior is caused by the potential barrier between the metal and semiconductor. Rectifying contacts have a current to voltage curve that is nonlinear and asymmetric about the origin. Usually, they have a very low resistance in the forward bias and an infinite resistance in the reverse bias. When a bias is applied, the semiconduc tor barrier will vary a nd the direction will change with the bias polarity while the me tal component of the barrier will remain constant. The dominant transport mechanism for a Sc hottky diode at room temperature is thermionic emission [12,33]. The majority carrier will move from the semiconductor
11 over the potential barrier into the metal. The current density based on the thermionic emission is JAT q kT qV kTB 21 exp()(exp()) (2-5) where A is the effective Richardson constant (A/K2-cm2), T is the absolute temperature, B is the barrier height, k is the Boltzmann constant, q is the electron charge, and V is the bias. 2.3.2 Ohmic Contacts When the semiconductor contact is ohmic, the carriers can flow in and out of the semiconductor freely with a minimum cont act resistance according to OhmÂ’s law [31,33,34]. The current to voltage characterist ics of the device are linear and symmetric. To make an ohmic contact on an n-type se miconductor, the work function of the metal must be less than or equal to the semiconduc tor electron affinity. This will decrease the barrier, allowing the carriers to flow freely, as shown in Figure 2-2. The barrier height for an n-type semiconductor is B = MS (2-6) where S is the semiconductor electron affinity , which is the difference between the vacuum level and the conduction band, and M is the work function of the metal. For an ohmic contact on a p-type semic onductor, the work function of the metal must be greater than or equal to the sum of the semiconductor electron affinity and the bandgap energy, B = Eg ( M S ) (2-7) where Eg is the bandgap. Then the bending of th e valence band in the semiconductor can be minimized and carriers can flow freely with a low barrier height , as shown in Figure
12 2-3. A high work function metal is a good ca ndidate for a p-type semiconductor while a low work function metal is the best candidate for an n-type semiconductor. However, there is a problem when considering wide ba ndgap materials for a p-type semiconductor. If the sum of Eg and s is greater than M , there will be a barrier to the electron flow. Consider the bandgap and elec tronic affinity of GaN and ZnO. By summing these values, which can be found in Table 2-2 [35, 36], a metal work function of 7.72 eV is required to create an ohmic contact. The me tal with the highest work function is Pt, which has the value of 5.64 eV . To complicate matters, there is almost always some level of contamination at the interface th at will increase the barrier height. Some pretreatment, such as etching, is required before depositing the metal. If a mismatch occurs between the fermi leve l of the metal and semiconducto r, a rectifying contact is formed. This means that no metal will be suitable for forming an ohmic contact on the semiconductor materials. It is not possible to find an appropriate metal to obtain an ohmic contact for most semiconductors. However, there are two additi onal ways to make an ohmic contact : the barrier height can be lowe red, or a tunnel contact can be created. The barrier height can be adjusted by alloying the metal with the semiconductor, or by changing the doping density. Annealing the sample will alloy th e metal with the semiconductor and reduce the barrier at the interface after the me tal deposition, hence improving the contact resistivity, but a tunnel contact is a more practical way to make an ohmic contact. Typically, this requires a doping density higher than 1019 cm-3 to make the barrier very thin. If the barrier region can be reduced to 3 nm or less, the carriers will be able to tunnel through this region.
13 To create effective metal c ontacts on semiconductors, either a suitable metal is required or a tunneling contact must be generated. Since the work function of most metals is less than 5 eV, there are great a dvantages in generating a tunnel contact on wide bandgap materials. In addition, it is critic al that the metal contact is thermally and electronically stable, has a low resistivity, and is non-reactive with a minimal tensile stress. 2.3.3 Transmission Line Method (TLM) The TLM was proposed by Shockley in 1964 . The contact resistance for a lateral current flow geometry can be obtained using the TLM, where a series of the samesized contacts with different gap lengths ar e deposited on the semiconductor, as shown in Figure 2-4 . Probes are placed at the two en ds of the pattern and are used to maintain a constant current throughout the measuremen ts. A second set of probes is used to measure the voltages between adjacent electrod es. Then the resistance between these two pads is calculated. The resistance versus th e gap between the pads is shown in Figure 25. The total resistance be tween two pads is given by RR R W lc s 2 (2-8) where Rc ( ) is the contact resistance, Rs ( / ) is the sheet resistance, l is the gap between the two pads, and W is the width of the pad. The slope of the resulting line is a function of the sheet resistance and the intercept is the contac t resistance. As the distance approaches 0, Rs goes to 0 and the total resistance becomes equal to 2 Rc. In general, a good contact resistivity is on the order of 10-7 -cm2 . The specific contact resistivity ( -cm2) is
14 c T s c sR R Rw R 2 2() (2-9) where RT ( -mm) is the transfer resistivit y. The minimum contact width W should be larger than the transfer length LT, which is expressed as L RT c s (2-10) LT provides a good measure of the contact quality. Ideally, LT should be targeted between tens of Âµm to several Âµm. A disadvantage of the TLM technique is that current crowdi ng may occur during the measurements . To address this issu e, similar measurements can be performed on the mesas shown in Figure 2-6, which can c onfine the current in the mesa pattern and reduce the current errors. The measurements can also be performed using a circular TLM (c-TLM), shown in Figure 2-7, which does not use an etching step to form a mesa and thus requires only one lithography step . For c-TLM, two probes are utili zed to measure the resistance across the structure. The resistance R is  R RR Rs L RsRsh T 2 112 222[ln()] (2-11) where Rsh is the sheet resistan ce of the semiconductor, R2 is the outer radius of the circular contact, s is the gap spacing, and LT is the transfer length. The measured resistance will yield a nonlinear resistance versus gap curve. To obtain the linear curve, and hence the contact resistance and sheet resistance, the following transformation is required. Assuming that R1 >> s and R2 = R1 + s, the resistance is
15 R R R sLcsh T2 21()  (2-12) where c is a correction fa ctor described by c R s Rs Rl ln 1 1 (2-13) Since the outer ring radius to gap ratio is larg er, the ring contact geometry reduces to that of the TLM model. 2.4 Bandgap Engineering To successfully generate an LED, knowledge of the bandgap, valence band offset, and conduction band offset is required. Th e bandgap will determine the wavelength of light emitted from the device. The electron flow is determined by the conduction band, and the hole movement is dictated by the va lence band. When creating heterostructures, the carriers (holes and electrons) need to be confined at the quantum well. At a heterojunction, the valence band and conductio n band deviate from their respective bulk material values. The observed change in th e offset values is a function of the bonding strain at the interface between the two materials. Ther e are three types of band lineup: (1) straddling, (2) staggered, and (3) broken-gap, as show n in Figure 2-8 . The energy bandgap will also increase with increasing temperature, as given by ETE aT Tgg 02 (2-14) where the Eg is the bandgap, T is the temperature, a is the first bandgap correction factor, and is the second bandgap corr ection factor. Electron spectroscopy can measure the binding energy difference between a semiconducto r core level and th e valence band edge of the semiconductor interface. Therefore, it is critical to understand the valence and
16 conduction band offsets when the heterojunc tion is formed at relevant operating temperatures (~ 25Â°C). This provides info rmation regarding the band bending, Schottky barrier height, and heterojunction band disc ontinuities [43,44]. The binding energy is measured with respect to the fermi level. X-ray photoelectron spectroscopy provides a direct contactless and non-destructive method to measure the binding energy with respect to the fermi level. With this information, the valence band offset can be determined from EEEEEEVCL a VBM a CL b VBM b CL()() (2-15) where Ev is valence band discontinuity, ECL a and ECL b are the core level binding energies for the bulk a and b samples, EVBM a and EVBM b are the valence band maximum binding energies for the bulk a and b samples, and ECL is the core level binding energy difference, as shown in Figure 2-9. The band offset at various semiconductor heterojunctions has been studied and es tablished by many groups [35,43-66]. The conduction band offset, Ec, is EEEcgv (2-16) where Eg is the bandgap difference between samples a and b. Only the binding energy difference is used to determine Ev. It is important to note that the EF location is not required to measure the heterojunction discont inuity. The core level peak position is defined as the middle of the full width at half maximum. The EVBM is determined by extrapolating from the leading edge of the valence band spectrum, as shown in Figure 210.
17 2.5 Processing Techniques 2.5.1 E-Beam Evaporator In an e-beam evaporator, the high-en ergy electron beam em itted from a hot filament is accelerated to a voltage of 10 keV [28,34]. The beam is then directed toward the target to be evaporated. The material heats up and be gins to evaporate as it is bombarded by the electron beam. To grow the f ilm, a substrate is placed in line with the e-beam target. Typically, the sample vapor pr essure should be at least 10 mtorr to obtain reasonable deposition rates. The evaporator mu st be operated at low rates to achieve the best uniformity; however, care must be ta ken since low evaporation rates require extremely high vacuums to minimize film contamination. 2.5.2 Electrical Measurements The I-V curves used to characterize the semiconductor devices are generated using a Hewlett Packard 4156A Precision Semiconducto r Parameter Analyzer. A probe station is connected to the analy zer to create a contact wi th the microscopic device. A four-point probe is used to measure th e resistivity of the semiconductor material . A constant current is applied to the tw o outer probes while the voltage is measured across the two inner probes. This provides the information required to calculate the resistivity. See Section 2.3. 3 for a more detailed description of this technique. 2.5.3 Rapid Thermal Annealing Rapid thermal annealing is used to rapi dly heat and cool the wafer. This minimizes the effects of particle diffusion [ 28,67]. Wafers are ann ealed to migrate the dopants to the desired depth by varying the temperature and annealing time. Figure 2-11 depicts a typical temperature ramp block. Ini tially, the sample is maintained at constant temperature while the chamber is purged with an inert gas. This will minimize any
18 atmospheric contamination before the annealing. The temperature is then rapidly ramped up to the annealing temperature. Care is taken to minimize the time during the ramp up for better control over dopant diffusion. Upon completion of the annealing, the wafer is rapidly cooled to terminate the dopant diffusion. The chamber is purged with an inert gas while the sample is cooled to minimize the gas-solid interactions while the sample is at elevated temperatures. The steady-state porti on of Figure 2-11 is governed by the length of time required to optimize the dopant distributions within the sample. 2.5.4 Sputter Deposition Sputtering is a process in which atoms ar e removed from the surface of a solid by high-energy ion impacts [28,34]. The materi al removal is a product of the momentum exchange between the incoming ion and the ta rget material. A pl asma consisting of argon ions and electrons is i gnited between the so urce and the substrate. The target material is placed on the electrode with the voltage set to maximize the ion flux at the target. As the positively charged argon ions bombard the target surface, the target material is removed from the surface. These atoms impinge on the sample surface, forming a thin film. The sputter yield depends on the direction of the incident ions, target material, mass ratio of the target material to the bombarding ions, and energy of the bombarding ions. Sputter de position has multiple advantages over other techniques, including improved film uniformity and an enhanced ability to produce layers of compound materials and alloys. 2.6 Characterization Techniques 2.6.1 Auger Electron Spectroscopy (AES) AES is a technique that is used to mo nitor surface cleanliness and identify the elemental composition of the near-s urface region and small area depth
19 profiling [67-69]. The resolution of the surface scan is approximately 3 nm and the depth profiling ranges from 2 to 20 nm. AES uses a focused electron beam directed toward a surface. The high-energy electron beam re moves a core level electron from a surface element. Then an electron within the at om, known as the Auger electron, decays with a probability of transferring this energy to anot her electron. The Auger electron is ejected into a vacuum. Finally, the kinetic energy of the Auger electron is measured with a detector. Since each element has its own char acteristic electronic stru cture, it is possible to determine the elemental composition on the surface, except for hydrogen and helium. Depth information can be determined by using AES in conjunction with ion bombardment, providing el emental depth profiles. 2.6.2 Atomic Force Microscopy (AFM) AFM measures the topography of conductors , semiconductors, and insulators with a force probe located within a few Ã… of the surface [70,71]. To co llect the AFM data, a tip is rastered across a surface while th e tip movements normal to the surface are measured. This technique has a lateral resolution of 1 to 5 nm. AFM is typically used to obtain a three-dimensional surface image or determine the surface roughness and crystal grain size. As the tip is rastered over the surface, a feedback mechanism is employed to ensure that the piezo-e lectric motors maintain a constant tip force or height above the sample surface. The tip movements normal to the surface are digitally recorded and can be manipulated and displayed in three-dimens ions by a computer. There are three types of AFM mode, described below. (1) Contact mode. Here, the force between the probe tip and substrate is repulsive with a mean value of 10-9 N. The force is set by pushing a cantilever against the sample surface. The contact mode can obtain a higher atomic resolution
20 than the other modes, but it may damage a soft material due to excessive tracking forces applied from the probe on the sample. Unlike the other modes, frictional and adhesive forces will affect the image. (2) Non-contact mode. Here, the main inte raction force between the probe tip and the substrate is attractive due to van de r Waals force. The tip is 5 to 150 nm above the sample surface. The resoluti on in this mode is limited by the viscous interactions with the surrounding atmosphere. (3) Tapping mode. Here, the tip is brought in to contact with the surface, lifts off the surface, and moves to the next lo cation where it is again brought into contact with the surface. The tapping mode overcomes the problems that the other two modes have by alternately pl acing the tip in cont act with the surface to provide a high resolution and then lifting the tip off the surface to avoid dragging it and damaging the material. The interaction between the probe tip and subs trate as a function of the distance is shown in Figure 2-12. 2.6.3 Photoluminescence (PL) Photoluminescence (PL) is a nondestructive t echnique that is used to measure the physical and chemical properties of materials [ 70]. A light is directed toward the sample surface. If the impinging light has a higher energy than the bandgap, an electron hole pair will be generated. Light will be emitted as these electron hole pairs release energy in radiative recombination. The emitted light is collected by a lens, passed through an optical spectrometer, and is measured at a detector. The radiative recombination mechanism, known as photoluminescence, is depi cted in Figure 2-13. PL is typically
21 used to determine the bandgap of a semiconductor, identify impurities, and verify the quality of a sample. 2.6.4 Profilometry Stylus profilometry is used to measur e the sample thickness and topographical features including feature widths , island heights, and trench depths. The best resolution that can be obtained using this technique is 10 nm. It is typically used to measure features with length scales on the order of microns. During the measurement, a stylus directly contacts the surface and is translated across the surface in one dimension. The vertical displacement of the stylus is convert ed to electrical signals, which are recorded by a computer. Profilometry may damage samples made from soft material. 2.6.5 Scanning Electron Microscopy (SEM) SEM is used to obtain topographic images of the microstructural to nanostructural characteristics of a solid material, wi th a magnification range of 10 Â– 10,000 . An important feature of SEM is th e large depth of field, which offers the ability for the specimen image to have a three-dimensi onal appearance. The components of SEM include an electron lens system, an elect ron source, an electron collector, and photorecording cathode ray tubes. The resolu tion of a conventional electron source is 10 Â– 50 nm while that of a field emission el ectron source is 1 Â– 5 nm. The sample is irradiated with a focused electron beam and signals, such as secondary electrons, backscattered electrons, characteristic x-rays, and other photons of various energies, are collected from the interacti on of the electron beam with the sample. A good SEM image can be optimized by the beam parameters. The limiting sharpness and feature visibility of the images depends on four parameters: the electron probe size dp, electron probe current ip, electron convergence angle p, and electron beam accelerating voltage V0, as
22 shown in Figure 2-14. SEM provides in formation about the surface topography, crystallography, and composition. 2.6.6 X-Ray Photoelectron Spectroscopy (XPS) XPS was developed in the 1960s by Kai Siegbahn, who was awarded the Nobel Prize for Physics in 1981 [67,69,73] . Common X-ray sources are Mg K (1253.6 eV), Al K (1486.6 eV), and monochromatic Al K (1486.7 eV). The photons can penetrate samples on the order of 1 Â– 10 microns. Th e XPS emission process is shown in Figure 215. In XPS, X-rays bombard a sample, whic h causes a core electron to be ejected from an atom. The electron then moves to the su rface and finally escapes into the vacuum. The elements on the sample can be identified from the binding energies of the ejected photoelectrons. The photoelectr on energies and intensities pr ovide information regarding the relative concentration and chemical state of the elements present on the sample. The only chemical species that ca nnot be detected by XPS are hydrogen and helium. XPS can probe 30 Ã… of the sample surface. The kinetic energies of the emitted electron are KE = h Â– BE s (2-17) where h is the photon energy, BE is the binding energy of the atomic orbital from which the electron originates, and s is the spectrometer work function. XPS is a surfacesensitive technique for chemical analysis, and the length scale is set by the inelastic mean free path in the solid. This technique is also used to determin e the conduction and valence band offset at heterojunctions, as described in Section 2.4.
23 Table 2-1. Physical properties of ZnO Property ZnO Crystal Wurtzite Bandgap 3.37 eV (RT), 3.44 eV (2K) Density 5.606 g/cm3 Space group P63mc Lattice constant a = 3.250Ã…, c = 5.204Ã… Sublimation point 1975 oC Thermal conductivity 0.6, 1.0 Â– 1.2 Thermal expanasion coefficient11 = 4.0, 33 = 2.1 (10-6 / oC) Refraction index no = 1.9985, ne = 2.0147 ( = 6328 Ã…); Effective mass me* = 0.28 m0, mhh* = 0.78 m0 Intrinsic carrier concentration < 106 cm-3 max n-type doping > 1020 cm-3 electrons max p-type doping < 1017 cm-3 holes Electron Hall mobility at 300K for low n-type conductivity 200 cm2/Vs (RT) Hole Hall mobility at 300K for low p-type conductivity 5 Â– 50 cm2/Vs Table 2-2. Bandgap and electr on affinity of GaN and ZnO. Material Eg (eV) (eV) Eg + (eV) GaN 3.4 4.1 7.5 ZnO 3.37 4.35 7.72
24 Zn O a c Zn O Zn O a c Figure 2-1. Crystal st ructure of ZnO.
25 Vacuum level EcEfEvcEgM metaln-type semiconductor Vacuum level EcEfEvcEgM metaln-type semiconductor Figure 2-2. Energy band diagram of a meta l/n-type semiconductor contact at thermal equilibrium.
26 Vacuum level EcEfEvcEgM metal p-type semiconductor Vacuum level EcEfEvcEgM metal p-type semiconductor Figure 2-3. Energy band diagram of a meta l/p-type semiconductor contact at thermal equilibrium. l1 l2l3 RcRcRsW l1 l2l3 RcRcRs l1 l2l3 RcRcRsW Figure 2-4. Schematic diagram of a TLM pattern.
27 R ( ) L 2LT 2Rc Slope =Rsheet/W R ( ) L 2LT 2Rc Slope =Rsheet/W Figure 2-5. Total resistance as a function of TLM pad spacing.
28 Metal pads Substrate Mesa Metal pads Substrate Mesa Figure 2-6. TLM without etch mesa (t op); TLM with etch mesa (bottom).
29 R2R1 s R2R1 s Figure 2-7. Top view of a circular TLM pattern. B A A B A B EcEv(a) (b)(c) B A B A A B A B A B EcEv A B EcEv(a) (b)(c) Figure 2-8. Energy band lineups : (a) straddling, (b) staggered, and (c) broken-gap.
30 EC aEV aEC bEV b EC EVabECL b (ECL bÂ–Ev b) Eg b Eg a ECL a (ECL aÂ–Ev a) ECLEF EC aEV aEC bEV b EC EVabECL b (ECL bÂ–Ev b) Eg b Eg a ECL a (ECL aÂ–Ev a) ECLEF Figure 2-9. Energy-band diagram at an abrupt heterojunction interface.
31 43210 EVBM Figure 2-10. Linear method applied to the spectrum results in the maximum valence band. DelayDelay Steady-State RampRamp DelayDelay Steady-State RampRamp Figure 2-11. Operating procedure of rapid thermal annealing.
32 Non-contact Contact Tapping modeRepulsive force Distance between sample and tip Attractive force Force Non-contact Contact Tapping modeRepulsive force Distance between sample and tip Attractive force Force Figure 2-12. Interaction between the probe tip and substrate as a function of distance.
33 Conduction band Valence band 1 2 34 5 EDEA Conduction band Valence band 1 2 34 5 EDEA Figure 2-13. Exciton generated because the in cident laser beam is greater than the bandgap. 1. Band to band recombination. 2. Electron to acceptor. 3. Donor to acceptor. 4. Donor to hole. 5. Other internal impurity of defect transitions. dp V0p ip dp V0p ip Figure 2-14. Four beam paramete rs of the SEM imaging mode.
34 Photon Photoelectron (1) (2) (3) (4) Auger electron K L L Photon Photoelectron (1) (2) (3) (4) Auger electron K L L Figure 2-15. XPS photoemission and Auger proc ess. (1) A photon interacts with the element. (2) The energy of the photon is channeled into a core electron, which departs from the atom. (3) The el ectron decays to the lower orbital, and the energy is transferred to another el ectron, called the A uger electron. (4) The Auger electron escapes from the element.
35 CHAPTER 3 WET ETCH OF ZINC OX IDE-BASED MATERIALS 3.1 Introduction ZnO has a long list of applications in gas sensing, in varistors, in piezoelectric transducers, as a light transmitting electrode in optoelectronic devices, in electro-optic modulators, and as a sunscreen . R ecently, major advances in the areas of conductivity control of epitaxi al films and availability of high-quality bulk ZnO substrates have refocused attention on Zn O/ZnMgO/ZnCdO heterost rucutre system for use in ultraviolet light emitters and transparent electronics . ZnO can be grown at relatively low temperatures on cheap substrates such as glass, and it has a larger exciton binding energy (~60 meV) than GaN (~25 meV) (enabling the fabric ation of vertical geometry devices with low threading dislocati on densities). The latter is important for achieving robust light emission at higher temper ature. The ZnO system also has simpler processing relative to GaN for which cannot be wet-etched in conventional acid mixtures at safe temperatures [21,59,75-92]. There ha ve been a number of recent reports of ZnObased light emitting structures involving different approach es, including homojunctions with low emission intensity [35,83,91-94]; hybrid structures involving heterostructures with SiC, GaN/sapphire, Al GaN/ sapphire, or Cu2O [35,93-97]; and ZnO pin homojunction diodes grown on oxide substrates [15,16,98]. Another important factor in the design of ZnO-based light-emitting diodes (LED) is the realization of bandgap engineering to create barrier layers and quantum wells in he terostructures. With respect to higher bandgaps, an increase up to 4.0 eV has been achieved by the incorporation of
36 Mg in the ZnO layer while still maintaining th e wurtzite structure . Ternary ZnCdO seems to be an appropriate candidate for narrow bandgap because of the smaller bandgap of CdO (2.3 eV), providing, of course, that phase separation is avoi ded during growth of this ternary . Molecular beam epitaxy (MBE) provides an excellent vehicle for controlling both purity and phase of the ZnCdO. In addition, the advances in growth of ZnO nanorods and nanowires have suggested th at these may have applications to biodetection  because of th eir large surface areas. In most device fabrication schemes, wet et ching is needed for isolation or mesa formation if nonconductive substrates are used . The binary ZnO is readily etched in many acid solutions, including HNO3/HCl and HF [89,98,100,101]. In most cases, the etching is reaction limited, with typical thermal activation energies of >6 kcal mol-1. In preliminary work, we have found that the et ching of the ZnO is strongly dependent on material quality. If the ZnO is very thin, the wet etch rates are high in all acid solutions. A particular problem encount ered with the wet etchi ng of ZnO/AlGaN-based LED structures was the presence of a very significant undercut (as much as around 10 Âµ m) , which occurred mainly at the end of the selective removal of the ZnO from the underlying AlGaN. There are no reports to date on the we t etching of ZnCdO and ZnMgO and in particular it is important to develop selective etches for ZnCdO/ZnO and ZnMgO/ZnO systems. We demonstrated the achievement of selective etching of ZnCdO over ZnO using both dilute HCl and H3PO4 mixtures. We also report on the selective etching of Zn0.9Mg0.1O relative to ZnO, both grown with sim ilar thicknesses on sapphire substrates by pulsed laser deposition (PLD) to ensure similar crystal quality. In Zn0.9Mg0.1O/ZnO
37 system, wet etch selectivities over 400 for ZnMgO over ZnO can be achieved with HCl at high dilution factors with water. 3.2 Experimental 3.2.1 Growth of ZnCdO MBE growth of ZnO and ZnCdO single-layers was performed on c-plane sapphire substrates. Low-temperature Knudsen cells were used to evaporate 6 N purity Zn, Cd, and Mg. Atomic oxygen fluxes were produced by a radio-frequency (rf) plasma source. The source was operated at an O2 flow rate of 2-5 standard cubic centimeters per minute, depending on the Zn flux, and a forward power of 150 W. In situ reflection high-energy diffraction (RHEED) was used to monitor MB E growth of the wurtzite phase ZnO and Zn0.95Cd0.05O. After thermal cleaning of the sapphi re surface, the s ubstrate temperature was lowered to 400Â–600Â°C. The ZnO growth was initiated by simultaneously opening the Zn and the atomic oxygen shutters to ensu re Zn-polarity for as-grown films. To reduce the number of variable growth parameters, TS (400Â°C) and Zn/O flux ratio (1.5:1) were kept constant and only the Cd cell te mperature was used to control the composition during growth. After growth, room temperature cathodoluminescence (CL) measurements were performed on the samples to define the spectral position of the nearbandedge emission peaks for ZnCdO. Th e Cd composition was verified by both Rutherford backscattering and the calibrated XPS measurements and was determined to be 4.8 Â± 0.3%. The experimental bandga p of our ZnCdO was ~2.9 eV from CL measurements at 25Â°C. There was no phase separation in the ZnCdO films, as confirmed by CL emission mapping.
38 3.2.2 Growth of ZnMgO The films were grown by PLD on c-plane Al2O3 substrates. Phosphorus-doped (Zn0.9Mg0.1)O targets were fabricated usi ng high-purity ZnO (99.9995%) and MgO (99.998%), with P2O5 (99.998%) serving as the doping agen t. The ablation targets were fabricated with a phosphorus doping level of 2 at.% (which can produce p-type conductivity if the films are annealed after grow th ), and a KrF excimer laser with a repetition rate of 1 Hz and pulse energy density of 1Â–3 J/cm2 was used as the ablation source. The target to substrate distance wa s 4 cm. The ZnO growth chamber exhibits a base pressure of 10-6 torr. Film growth was performed at 500Â°C for ZnMgO or 400Â°C for ZnO in an oxygen pressure of 50Â–150 mTorr. Both types of films were ~0.5-mm thick with an electron concentration of 1017 cm-3 and mobility of 25 cm2Â·V-1Â·s-1 for the ZnO and a mixed conductivity (indeterminate conducti vity type with carrier density around 1016 cm-3) for the ZnMgO layers, as determined fr om van der Pauw Hall measurements. 3.2.3 Wet Etch Method The wet chemical etch was performed with HCl/H2O or H3PO4/H2O solutions as a function of both solution concentration and temperature. A photor esist mask (AZ 1045) was used for creating features whose depth was measured by stylus profilometry after postetch removal of the mask in acetone. 3.3 Results and Discussion Figure 3-1 and 3-2 shows the etch rates of Zn0.95Cd0.05O and Zn0.9Mg0.1O, respectively, as a function of solution concentration for HCl/H2O or H3PO4/H2O at 25Â°C. For Zn0.95Cd0.05O, controllable etch rate s in the range (<100 nmÂ·min 1) are desirable for mesa formation and were obtained over this set of solution concentrations. For Zn0.9Mg0.1O, the etch rates were sign ificantly faster with HCl/H2O at all concentrations
39 (120Â–1100 nmÂ·min-1). Note, high dilution factors of th e acids with water were used to obtain controllable etch rates. The use of pure HCl or H3PO4 produced very high rates and extensive bubbling in the solutions th at led to nonuniform and rough surfaces. To determine the rate-limiting step in the et ching, we measured the etch rate as a function of solution temperatur e over the range 25Â–75Â°C. In the dilute mixtures used here, it is common to have an etch rate of semiconductor is limited by the diffusion of the active etchant species to the ZnCdO and Z nMgO surface, or by the out-diffusion of the soluble product, i.e., a diffusion-limited etch ra te. Further characteristics include a square root dependence of etch depth on etch time, an activation energy 6 kcalÂ·mol 1, and a strong dependence of etch rate on solution agitation. This mode of etching is not desirable for device fabrication because of the difficulty in obtaining re producible rates. Figure 3-3 and 3-4 shows an Arrhenius plot of ZnCdO (~310 3 M) and ZnMgO (2.410 2 M) etch rate in the two solutions of HCl and H3PO4 at high dilution factors with water, respectively. Under these conditi ons, the etch activation energies are in the range ~0.4 kcalÂ·mol 1 for ZnCdO and 2-3 kcalÂ·mol 1 for ZnMgO, values that are consistent with diffusion-limited etching. Figure 3-5 shows the etch rates of ZnO as a function of solution concentration for HCl/H2O or H3PO4/H2O at 25Â°C. By contrast to the case of ZnMgO, the etch rates are significantly faster with H3PO4/H2O at all concentrations. Note that lower acid dilution factors were used to obtain controllable ZnO et ch rates, when compared to those used for ZnMgO. Once again, the use of pure HCl or H3PO4 produced irreproducible etch rates and nonuniform, rough surfaces.
40 The etch rates for ZnO were si gnificantly slower than ZnCdO (210 3-3.510 3 M) and ZnMgO (510 3-2.510 2 M) under the same conditions, and we had to employ lower dilution factors (210 3-10 1 M) in order to get practical removal rates. Figure 3-6 shows an Arrhenius plot of ZnO etch rate in the two acid/water mixtures. Under these conditions the activation energies (~5.6Â–5.9 Â± 0.4 kcalÂ·mol 1) were consistent with a transition to reaction-limited etching. Also th e transport of the reaction species through the etch solution is less of a factor than the very dilute solutions used for ZnCdO and ZnMgO, as the etch rate did not vary with changes in the so lution agitation rate. For the reaction-limited etch mechanism, the etch de pth depends linearly on time, the activation energy is 6 kcalÂ·mol 1, and the rate is independent of solution agitation. This is the preferred mode of etching for device fabr ication, as the temperature and solution composition need to be controlled, without the influence of varying composition gradients throughout the mixture. The resulting selectivities for etching ZnCdO over ZnO and ZnMgO over ZnO in the two mixtures with the same concentr ation are shown in Figure 3-7 and 3-8, respectively. The ZnO etch ra tes were in the range 30Â–90 nmÂ·min 1 for dilute acid mixtures. For ZnCdO, the HCl/H2O solution provides selectiv ities in excess of 50 under optimum conditions, while the maximum selectivity with H3PO4/H2O was ~15. ZnMgO had selectivities in exce ss of 425 with the HCl/H2O solution, under optimum conditions. A rule of thumb in device fabrication schemes is that a selectivity of at least 10, and preferably 100. Optical microscopic images were taken after selectively etching ZnCdO layers from an underlying ZnO layer on sapphire and ZnMgO layers from an underlying ZnO
41 layer on sapphire. These clea rly demonstrate that it is possible to generate a clean pattern transfer using selec tive etching. For example see the optical microscopy image and the scanning electron microscopy images of ZnCdO selectively etched with HCl solution from an underlying ZnO layer, as s hown in Figure 3-9. Similar results are shown for ZnMgO selectively etched from an underlying ZnO layer in Figure 3-10. 3.4 Summary ZnCdO, ZnMgO, and ZnO can be readily et ched in dilute solutions of HCl and H3PO4. High dilution factors of these acids with water provides controllable etch rates in the range 30Â–90 nmÂ·min 1 for ZnCdO and 120Â–1100 nmÂ·min-1 for ZnMgO, with adequate selectivity to ZnO grown under the same cond itions. Photoresist pr ovides a stable and convenient mask for patterning ZnCdO, ZnMgO , and ZnO in these acid solutions. The availability of simple wet solutions for this heterostructure system simplifies the processing of mesa-type ZnObased LEDs and avoids the need for plasma etching processes which are known to damage the ZnO surface even at low plasma powers .
42 0.00150.00200.00250.00300.00350.0040 30 40 50 60 70 80 90 100 HCl H3PO4RT Etch Rate (nm/min)Concentration (M) Figure 3-1. Etch rate of ZnCdO with different concentra tions of HCl and H3PO4 solutions diluted in water.
43 0.0050.0100.0150.0200.025 0 250 500 750 1000 HCl H3PO4 Etch rate (nm/min)Concentration (M) Figure 3-2. Etch rate of Zn0.9Mg0.1O in different concentr ations of HCl and H3PO4 diluted with water.
44 2.82.93.03.188.8.131.52 4 5 6 0.0031M HCl, Ea=0.37 Kcal/mol 0.0029M H3PO4, Ea=0.38 Kcal/mol Etch rate (nm/min)1000/T(K-1) Figure 3-3. Arrhenius plot of ZnCd O etch rate in 0.0031M HCl and 0.0029M H3PO4 solutions diluted in water.
45 184.108.40.206.03.13.23.33.43.5 1022x1023x1024x1025x1026x102 0.024M H 3 PO 4 E a = 2.07 Kcal/mol 0.024M HCl Ea= 3.29 Kcal/mol Etch rate (nm/min)1000/T (K -1 ) Figure 3-4. Arrhenius plot of Zn0.9Mg0.1O etch rate in 0.024M concentrations of HCl and H3PO4 diluted with water.
46 0.00.20.40.60.81.01.2 0 10 20 30 40 50 HCl H3PO4 ZnO substrate Etch rate (nm/min)Concentration (M) Figure 3-5. Etch rate of ZnO in di fferent concentrations of HCl and H3PO4 diluted with water.
47 2.82.93.03.220.127.116.11 10 100 1000 10000 E a = 5.88 Kcal/mol ZnO substrate 0.24M HCl Etch rate (nm/min)1000/T (K-1)E a =5.59 Kcal/mol ZnO substrate 0.060M H 3 PO 4 Figure 3-6. Arrhenius plot of ZnO etch rate in 0.24M HCl and 0.06M H3PO4 diluted with water.
48 0.00200.00250.00300.0035 0 20 40 60 HCl H3PO4 Etch selecivityConcentration (M) Figure 3-7. Etch selectivity of ZnCdO to ZnO at room temperat ure as a function of solution concentration.
49 0.0100.0150.0200.0250.030 0 50 100 150 200 250 300 350 400 450 HCl H3PO4 Etch selecivityConcentration (M) Figure 3-8. Etch selectivity of ZnMgO to Zn O as a function of solution concentration.
50 Figure 3-9. Optical microscopy image (top) of ZnCdO selectively etched with HCl solution from an underlying ZnO layer. The photoresist mask is still in place. At bottom is an SEM image of the se lective removal of ZnCdO from an underlying ZnO substrate.
51 a b Figure 3-10. SEM images of ZnMgO selectivel y etched with an underlying ZnO layer. (a) The overall pattern w ith a 3000 X mag. (b) a 3700 X mag image showing the ZnMgO sidewall.
52 CHAPTER 4 OHMIC CONTACTS TO ZINC OXIDE-BASED MATERIALS 4.1 Introduction Zinc oxide (ZnO) exhibits an interesting combination of multifunctional properties, including optical, semiconducting, piezoel ectric, electrooptic, and optoelectronic functionality, and ZnO films find applicati ons in many electroni c devices including sensors, actuators, transducers, and high frequency surface acoustic wave (SAW) devices . There are also uses for n-ZnO in the copper indium gallium diselenide (CIGS, or Cu(InxGa1-x)Se2) thin-film solar cell, where there is strong interest in ZnO (and its alloys) to solid-state lighting. ZnO devices may ha ve application in th e photoelectrochemical splitting of water to generate hydrogen . There are also some significant recent interests in developing ZnO-based UV light-em itting diodes (LEDs) that may have some advantages over the corresponding devices fabric ated in the GaN material system. These include the lower deposition temperatures for ZnO, the higher exciton binding energy (important for robust light emission over a broader range of temperatures), the commercial availability of high quality substr ates and the simpler processing relative to GaN for which convenient wet etch are not available [19,21,109]. ZnO has a bandgap of about 3.2 eV and its band-edge emission is about 388 nm, in the bl ue/near-ultraviolet portion of the electromagnetic sp ectrum. If efficient lasers or LEDs can be made of ZnO, the near-ultraviolet photons that would be emitted could be used to excite phosphors in suitable combinations to produce white light . Alternatively, alloying with cadmium would reduce the bandgap, and the emitted photons would be of longer wavelength .
53 There have been many recent advances in im proved ZnO epitaxial growth quality and development of the various processing modules necessary for LED fabrication. A variety of ZnO-based LEDs, including homojunctions with low emission intensity [83,91,92], hybrid structures involving heterostructures with SiC, GaN/sapphire , AlGaN/ sapphire or Cu2O [35,93-97], and ZnO p-i-n hom ojunction diodes grown on ScMgAlO4 or Al2O3 substrates have been reported recently [15,16]. Another important factor in the design of ZnO-based LED is the realization of bandgap engineering to create ba rrier layers and quantum wells in heterostructures. An increase up to 4.0 eV has been achieved by the incorporation of Mg in the ZnO layer while still maintaining the wurtzite structure[ 98]. On the other hand, ZnCdO is the best candidate for the narrow bandgap component of a heterostructure because of the smaller bandgap of CdO (2.3 eV) . However, very little has been reported on development of ohmic contacts to ZnCdO. In ZnO, the usual approaches invo lve surface cleaning to reduce barrier height or increase of the e ffective carrier concen tration of the surface through preferential ion of oxygen [8-10,78,91,111-125]. Specifi c contact resistances of ~ 310-4 -cm2 were reported for Pt-Ga contacts on n-ZnO epitaxial layers [99,112], 210-4 -cm2 for Ti/Au on Al-doped epit axial layers [113,114], 0.7 -cm2 for nonalloyed In on laser-processe d n-ZnO substrates , 2.510-5 -cm2 for non-alloyed Al on epitaxial n-type ZnO  7.310-3 -cm2 to 4.310-5 -cm2 for Ti/Au on plasma exposed, Al-doped n-type epitaxial ZnO and 910-7 -cm2 for Ti/Al on n+-epitaxial ZnO . Several points are clea r from the past works, namely that the minimum contact resistance generally occurs for post-de position annealing te mperatures of 200C to 300C on doped samples treated so as to further increas e the near-surface carrier concentration.
54 A necessary component of any ZnO-based light emitter is low resistance ohmic contacts to both nand p-type layers which are thermally stable and reliable. There have not been many detailed studies of contact fo rmation to ZnO with controlled intentional doping levels [9,10,115-117,126-128]. The usual approaches so far have focused on (i) use of various surface preparations to redu ce the metal semiconductor barrier height or increase of the carrier tunneling probability, or by (ii) increasing the near-surface carrier concentration through intentiona l oxygen loss or the indiffusi on of zinc. A variety of different metallization schemes have been re ported for ohmic cont acts on n-type ZnO, including Ti/Au, Ti/Al/Pt/Au, Zn/Au, Al, Al/Pt, Re/Ti/Au, Ru and Pt/Ga. These metallizations produce specific contact resistances in the range 10-4-10-7 cm2 on unintentionally n-type ZnO. In most cases in device applications, it is necessary to intentionally dope the n-contact layer because of the effect of contact resistance on device heating and reliability. A key aspect of the design of ZnO-ba sed LEDs is the need for transparent conducting oxides that increase current spread ing and reduce re-absorption of the photons generated in the smaller bandgap ZnCdO activ e layer. In GaN LED technology, indiumtin-oxide (ITO) is used as an anti-reflective coating with a low refractive index of ~2 at wavelengths in the range 400-460nm [129-134]. This compares to the refrac tive index of ~2.4 for ZnO at the bandgap wavelength. Prev ious results have shown that the optical transmittance of ITO films can be >85% over a fairly broad range of wavelengths (400520 nm) even after annealing to improve c ontact resistance [130,132]. The ITO is usually not conducting enough to make a direct oh mic contact, but is used as an overlayer on thin metal contact stacks [129-134]. For n-ZnO, specific contact resistances of 710-
55 3-4.310-5 -cm2 have been reported for Ti/A u on Al-doped epitaxial layers [113,114,116]. Generally, the minimum cont act resistance generally occurs for postdeposition annealing temperatures of 200 C to 300 C on doped samples. The thermal stability of most ohmic contact schemes on Zn O is far lower than those on GaN due to the lower thermodynamic stabili ty of most oxide phases re lative to their nitride counterparts. However, no work has been done to incorporate ITO onto the metal layers on ZnO and determine the thermal stability of the resulting stacks. In this Chapter we report an investigation of Ti/Au and Ti/Al/Pt/Au metallization on n-type ZnCdO layers grown on ZnO/GaN/Sapphi re templates. The effect of thermal annealing on contact resistance, contact mo rphology and stability were determined for annealing temperatures up to 600Â°C. We al so describe the prope rties of Ti/Au ohmic contacts to heavily Al-doped ZnO grown by Pulsed Laser Deposition (PLD). The contacts are Ohmic even in the as-deposited state due to carrier tunneling and additional annealing further reduces the sp ecific contact resistance. Th en, we investigate ITO/Ti/Au metallization on n-type ZnO:Al on sapphire temp lates. The effect of thermal annealing on contact resistance, contact morphology and stability were determined for annealing temperatures up to 450Â°C. The typical resistivity of the ITO film was around 10-3 ~ 104 -cm and the resulting contact resistivity was <10-5 Â·cm2 for all temperatures up to 450Â°C. 4.2 Experimental 4.2.1 Ti/Au and Ti/Al/Pt/Au Ohmic Contacts to N-type ZnCdO MBE growth of ZnO and ZnCdO single layers was performed first on GaN/c-plane sapphire templates to optimize growth conditi ons. The GaN templates were transferred
56 to an oxide MBE system and were outgassed at ~700C for 15 min under N2 ambient to obtain a clean surface, as indicated by a sh arp reflection high-ene rgy electron diffraction (RHEED) pattern. The in-situ RHEED was us ed to monitor MBE growth of ZnO and ZnCdO. After thermal cleaning of the Ga N surface, the substrate temperature was lowered to 400-600C. The ZnO growth was initiated by simultaneously opening the Zn and the atomic oxygen shutters to ensure Zn-p olarity for as-grown films. During growth, the oxygen flow rate was controlled at 2-5 s ccm flow rate, depending on the Zn flux. Under these conditions ZnO growth rates of 0.4-1 m/hr were obtained depending on the substrate temperature. After 1 hour of growth, the thicknes s of the ZnO film was about 0.6 m and a spotty RHEED pattern was observed. The RHEED pattern suggests that the ZnO film is single crystal grown along the  direction, or c-axis, without in-plane rotation. However, the spotty nature of the pattern indicates that the film is not atomically flat. After growth, th e samples were annealed to ~700C in vacuum (10-5 Torr). This step was found to improve the rough ness of the films. Based on the results of CL measurements, the highest-quali ty ZnO films can be grown at 400C with a Zn/O flux ratio of approximately 1:1 on clean GaN surf ace without any special prior treatments. Further improvements in ZnO surface mor phology and electrical properties were obtained by annealing at 700C in vacuum for 1 h. The thicknesses of the ZnCdO films grown in these experiments we re in the range of 0.3-0.8 m. In-situ RHEED was used to monitor MBE growth of the wurtzite phase ZnO and Zn0.95Cd0.05O. After thermal cleaning of the sapphire surface, the subs trate temperature was lowered to 400-600C. The ZnO growth was initiated by simultaneously opening the Zn and the atomic oxygen shutters to ensure Zn-polarity for as-grown films. To reduce the number of variable
57 growth parameters, TS (400Â°C) and Zn/O flux ratio (1.5 :1) were kept constant and only the Cd cell temperature was used to cont rol the composition during growth. The Cd composition was verified by both Rutherford backscattering and the calibrated XPS measurements and was determined to be 4. 8+/-0.3%. The experimental bandgap of our ZnCdO was ~ 2.9 eV from cathodoluminescence (C L) measurements at 25Â°C. There was no phase separation in the CdZnO films, as confirmed by CL emission mapping. Contact metallurgy of eith er Ti(200 Ã…)/Au(800 Ã…) or Ti(200 Ã…)/Al(800 Ã…)/Pt (400 Ã…)/Au (800 Ã…) was deposited by electron-beam evaporation and patterned by resist liftoff to form a transmission line pattern (TLM), consisting of 100 Âµ m2 and gap spacings varying from 2 Âµ m to 16 Âµ m. The samples were annealed for 1 min at temperatures up to 600Â°C in N2 and the current-voltage (I-V) characte ristics recorded on an Agilent 4156B parameter analyzer. Auger Electron Spectro scopy (AES) depth profiling was performed on a Physical Electronics 660 Scanning A uger Microprobe. The electron beam conditions were 10 keV, 0.3 Âµ A beam current at 30Â° from sample normal. For depth profiling, the ion beam conditions were 3 keV Ar+, 2.0 A, 3 mm2 raster, with measured sputter rate of 110 Ã…/minute. Cross-sect ion transmission electron microscopy (TEM) was also performed on the as-grown samples. 4.2.2 Ti/Au to Aluminum Doped ZnO The films were grown by PLD on MgO buffers on c-plane Al2O3 substrates. Figure 4-1 shows a cross-sectional TEM image of the as-grown sample. The deposited layers show a typical dislocation density for hetero epitaxial growth on a mismatched substrate and are typical of state-of-t he-art ZnO grown on sapphire. The ablation targets were fabricated with an Al doping level of 0.01 at. % and a KrF excimer laser with a repetition
58 rate of 1 Hz and pulse energy density of ~3 J/cm2 was used as the ablation source. The target to substrate distance was 4 cm. The ZnO growth chamber exhibits a base pressure of 10Â–6 Torr. Film growth was performe d at 800Â°C in an oxygen pressure of 20Â–50 mTorr. The films were ~ 0.9-Âµm thick with an electron concentration of 91018-1.31019 cmÂ–3 and mobility 50-61 cm2Â·VÂ–1Â·sÂ–1at room temperature, as determined from van der Pauw Hall measurements. The full-width-at-hal f maximum (FWHM) fo r the ZnO(002) is at the peak of 0.26-0.64 Â°. The layers exhibited strong band edge photoluminescence at ~370 nm at room temperature. Contact me tallurgy of either Ti (200Ã…)/Au (800Ã…) was deposited by electron-beam evaporation and patterned by resist lift-off to form a TLM with 100 Âµm pads spaced from 2 to 16 Âµ m. The samples were annealed for 1 min at temperatures up to 450Â°C in O2 and the I-V characteristics re corded on an Agilent 4156B parameter analyzer. AES depth profiling wa s performed on a Physic al Electronics 660 Scanning Auger Microprobe. The elect ron beam conditions were 10keV, 0.3 Âµ A beam current at 30Â° from sample normal. For de pth profiling, the ion b eam conditions were 3 keV Ar+, 2.0 A, 4 mm2 raster, with measured sputter ra te of 88 Ã…/minute. Finally, the surface roughness of the contacts was de termined by tapping mode Atomic Force Microscopy (AFM). 4.2.3 ITO/Ti/Au Ohmic Contacts to Aluminum Doped ZnO The films were grown by PLD on 200 nm thick MgO buffers on c-plane Al2O3 substrates. The ablation targ ets were fabricated with an Al doping level of 0.01 atomic % and a KrF excimer laser with a repetition rate of 1 Hz and pulse energy density of ~3 J/cm2 was used as the ablation source. The target to substrate distance was 4 cm. The ZnO growth chamber exhibits a base pressure of 10Â–6 Torr. Film growth was performed
59 at 800 Â°C in an oxygen pressure of 5 mTorr. The films were ~ 0.7 Âµm thick with an electron concentration of 1.31019 cmÂ–3 and mobility ~56 cm2Â·VÂ–1Â·sÂ–1at room temperature, as determined from van der Pauw Hall measurements. The FWHM for the ZnO(002) is at the peak of ~0.39 Â°. The layers exhi bited strong band edge photoluminescence at ~370 nm at room temperature. Contact metallur gy of Ti (200Ã…)/Au (800Ã…) was deposited by electron-beam evaporation and pa tterned by resist lift-off to form a circular transmission line pattern (c-TLM) with maximum outer pad contact radius of 160 Âµm. The ITO layers were deposited by sputtering from a target with composition 90% In2O3 10% SnO2. The deposition rates were in the range 0.4-0.6 Ã…Â·s-1, depending on power. The typical resistance of ITO film is around 10-3 ~ 10-4 -cm. The samples were annealed for 1 min at temperatures up to 450Â°C in O2 and the I-V characteristic s recorded on an Agilent 4156B parameter analyzer. AES depth profilin g was performed on a Physical Electronics 660 Scanning Auger Microprobe. The electron beam conditions were 10 keV, 0.3 Âµ A beam current at 30Â° from sample normal. For depth profiling, the ion beam conditions were 3 keV Ar+, 2.0A, 4mm2 raster, with measured sputter rate of 88 Ã…/minute. 4.3 Results and Discussion 4.3.1 Ti/Au and Ti/Al/Pt/Au Ohmic Contacts to N-type ZnCdO Figures 4-2 and 4-3 show TEM cross-sections of the layer structure. The layers show excellent interfacial abruptness and def ect densities typical of heteroepitaxial films grown a lattice-mismatched substrate. The crystalline quality of the samples was also investigated by selected area electron diffraction (SAED) patterns, as shown at the bottom of Figure 4-2. The SAED pattern taken across the interface of ZnO and GaN showed only one set of high-quality, well-define d single-crystal diffraction spots, without
60 any sign of spot broadening or splitting. The SAED pattern from the ZnO/GaN interface region looked essentially the same as th at obtained from the middle of the GaN layer. In contrast, the SAED pattern taken across the interface of GaN and sapphire, clearly shows splitting in diffraction spots, as expected for highly mismatched epitaxial growth. The as-deposited contacts were rectifying, but converted to Ohmic behavior after 200Â°C anneals and their properties improved st eadily up to 450-500Â°C. Figure 4-4 shows the sheet resistance under the ZnCdO derived from the TLM measurements, as a function of annealing temperature for both types of me tal schemes used. This decrease in sheet resistance corresponds to an improvement in both the contact transf er resistance (Figure 4-5) and specific contact resistance (F igure 4-6). The mini mum specific contact resistivity of 2.310-4 -cm2 was obtained at 500Â°C for Ti/Al/Pt/Au and 1.6 x10-4 -cm2 was obtained at 450Â°C for Ti/Al. These values also correspond to the minima in transfer resistance for the contacts. The mechanism of current tr ansport was elucidated by measuring the change in specific contact resistivity with the measurem ent temperature. Since the ZnCdO layer is moderately doped (~1017 cm-3), thermionic emission was expected, which is dependent upon the measurement temperature. Howeve r, the Ti/Au contacts exhibited almost constant specific contact resistance in the temperature range of 25-200oC, indicating that current flow is dominated by tunneling. Wh en the tunneling dominates, the specific contact resistivity (RSCR) is dependent upon doping con centration and is given as )] ( 2 exp[* D B e S SCRN m R (4-1)
61 where B is the barrier height, S the semiconductor permittivity (S,ZnO=8.5), * em the effective mass of electrons, the PlanckÂ’s constant and ND is the donor concentration in the semiconductor. It is reported that very thin titanium oxide layers can be formed at the interface when the Ti is cont acted to pure ZnO even in as -deposited condition  since the titanium has a higher affi nity with oxygen than Zn (Hf o (Ti3O5)= -2459.1 kJ/mol, Hf o (Ti2O3)= -1520.9 kJ/mol, Hf o (TiO2)= -944 kJ/mol, Hf o (TiO)= -542.7 kJ/mol vs. Hf o (ZnO)= -350 kJ/mol) . As a result , the oxygen vacancies, which are effective electron donors, increase carri er concentration near the ZnCdO surface promoting the tunneling phenomena through the extremely thin oxide barrier, as shown schematically in Figure 4-7. If Ti is brought into intimate contact with Zn CdO without any surface states and interfacial layers, the barrier height (B =Ti-ZnO) is calculated to be -0.02 eV based on electron affinity of ZnO (ZnO =4.35 eV)  and work function of Ti (Ti =4.33 eV), suggesting a negligible barrier height. The ba rrier height can be e xperimentally obtained from the measurement of specific contact resistivity (ln RSCR) vs. doping concentration (1/ND) in the field emission region [135,136]. From the results of as-deposited Ti/Al/Pt/Au contacts to n-ZnO films with a ra nge of carrier concentrations, the calculated barrier height is ~0.04 eV . The morphology of the contacts was a strong function of the annealing temperature, as shown in Figure 4-8. The smooth and fl at metal surface in the as-deposited condition for Ti/Al/Pt/Au became rough after 350oC annealing. After 600oC annealing, some voids were observed and the underlying ZnCdO laye r was exposed. As annealing temperature increases, Ti and Au start to form intermetallic compounds  and oxygen is predominantly removed from the ZnCdO surface leading to decomposition of the
62 ZnCdO. This combined effect appears to be responsible for the fo rmation of voids on the surface. By sharp contrast, the Ti/Au contac ts showed more superior thermal stability and still retained smooth morphology up to 450ÂºC. AES surface scans from the Ti/Au contacts ar e shown in Figure 4-9 as a function of annealing temperature. Zn and Ti outdiffusion to the surface are clearly evident by 450ÂºC. A summary of the near-surface compos ition data is shown in Table 4-1. The corresponding depth profiles are shown in Figure 4-10. The formation of the TiOx interfacial region is clearly ev ident after annealing and is st rong evidence that this is at least part of the reason for the improved contact resistance. Similar data is shown in Figures 4-11 and 4-12 for the Ti/Al/Pt/Au contacts on ZnCdO and a summary of the near-surface comp osition data is shown in Table 4-2. The surface scans show Au and Al outdiffusion to the surface are clearly evident by 450oC in the metallization scheme. The depth profiles show significant outdi ffusion of Pt, Al and Ti at the higher anneal temper atures and oxidation of the Ti , as discussed above. The Pt does serve as an effective diffusion barrier, at l east to 450ÂºC in this metallization scheme. 4.3.2 Ti/Au to Aluminum Doped ZnO Figure 4-13 shows the sheet resistance of the Al-doped ZnO under the Ti/Au contacts (top) and contact transfer resist ance (bottom) as a function of annealing temperature for Ti/Au contacts on Aldoped ZnO of two different electron concentrations. The latter were obtained by altering the oxygen pressure during growth of the ZnO. The sheet resistance shows little change with annealing temperature while the transfer resistance goes through a minimum between 200-300Â°C, depending on the doping concentration in the ZnO.
63 Figure 4-14 shows the resulting specific contact resistance of the Ti/Au contacts. The minimum specific contact resistivity of 6 x10-8 -cm2 was obtained at 300Â°C for Ti/Au, which is the lowest, reported for n-Ohmi c contacts on ZnO. No te that even the asdeposited contacts exhibit a very low specific contact resistivity of 2.410-7 -cm2 for the layer doped at 1.31019 cm-3. The specific contact resistiv ity of both the as-deposited and annealed contacts was independent of the measurement temperature over the range 25-225Â°C, the limit of our test set-up. This is characteristic of contacts in which the dominant transport mechanism is tunneling through the barrier. For the tunneling mechanism, the relation between specific contact resistivity (RSCR) and doping concentration ND is given by Equation 4-1. The high doping concentration in the ZnO promotes tunneling over the small barrier height of Ti (0.02-0.04 eV) on ZnO . In addition, it is thermodynamically favored that titanium oxide layers will form at the metal/ZnO interface even in as-deposited conditi on due to the stronger affinity of Ti for oxygen relative to Zn . If indeed th ere is formation of a thin titanium oxide interfacial layer, then the lo ss of oxygen from the ZnO would be expected to increase the carrier concentration near the ZnO surface and increase th e tunneling probability since oxygen vacancies are expected to be donors in ZnO. In our heavily n-type ZnO the additional donors are not re quired to create a low resistance contact. Figure 4-15 shows optical microscopy images of the Ti/Au contacts as a function of anneal temperature. Th e morphology of the contacts degraded with increasing temperature and was quite rough at 450oC. To quantify this effect, AFM scans were performed. Figure 4-16 shows topographic images of (left) as received Ti/Au deposited on ZnO:Al(N ~ 9.x1018 cm-3) and (right) after a subs equent anneal to 250 oC. The scan
64 size of these AFM images is 5 Âµm x 5 Âµm. The as-deposited contacts showed a rootmean-square (RMS) roughness of 43.2 nm while at 250oC annealing the RMS roughness was 40.7 nm. However, the as-deposited cont acts had the Z range of 223.3 nm with a smaller grain size and 250o anneal the Z range was 273.2 nm. Therefore, the surface did become rougher for anneals up to 450oC. However, this appeared to be somewhat sample-dependent, with the more heavily doped layers showing a lower degree of roughening at a given anneal temperature. The reason for this is not understood at present. As annealing temperature increases , Ti and Au start to form intermetallic compounds [10,117,119,127,128] and oxygen is predominantly removed from the ZnO surface leading to decomposition of the ZnO [126 ]. These reactions are likely to play a strong role in the roughening of the contact morphology but it is not clear as to why this should vary due to doping concentration in the ZnO. Figure 4-17 shows AES surface scans (top) a nd depth profiles (bottom) from the Ti/Au contacts both before and after anneali ng at 250ÂºC. Zn outdiffusion to the surface is clearly evident at 250ÂºC. The possible formation of a TiOx interfacial region is inconclusive from this data although we have seen more clear evidence in previous section on Ti/Au and Ti/Al/Pt/Au Ohmic contacts on n-type Zn0.05 Cd0.95O layers grown on ZnO buffer layers. 4.3.3 ITO/Ti/Au Ohmic Contacts to Aluminun Doped ZnO Figure 4-18 shows the specific contact resi stance of the ITO/Ti/Au contacts on Aldoped ZnO as a function of annealing temp erature, as well as the corresponding semiconductor sheet resistance under the cont act stack. The mini mum specific contact resistivity of 4.610-6 -cm2 was obtained at 50Â°C. Note that even the as-deposited
65 contacts exhibit a very low speci fic contact resistivity of 5.910-6 -cm2 and values of <10-5 -cm2 were achieved for all temperatures up to 450Â°C. This is attractive from the viewpoint of maintaining a low process th ermal budget for ZnO structures grown on cheap transparent substrates such as glass. The morphology of the contacts was a not a strong function of the annealing temperature until temperatures of 450Â°C, as shown in Figure 4-19. The as-deposited contact stack exhibited a smooth surface (a root-mean-square (RMS) roughness of ~5 nm as measured by Atomic Force Microscopy over a 55 Âµm2 area). The surface showed similar morphologies for anneals up to 450oC. At these annealing temperatures, previous results have shown that Ti and Au start to form intermetallic compounds [10,117,119,126-128] and oxygen is removed from the ZnO surface leading to decomposition of the ZnO . These re actions most likely play a role in the roughening of the contact morphology. Figure 4-20 shows the AES surface scans of ITO/Ti/Au contacts either asdeposited (bottom), after annealing 350oC (middle) or 450oC (top). Outdiffusion of In to the surface of the contact stack is evident by 350Â°C and Zn at the higher temperature. The corresponding elemental dept h profiles obtained by AES ar e shown in Figure 4-21. The Ti is an effective diffusion barrier to movement of In until 350Â°C annealing. At 450Â°C, the interfacial abruptness degrades due to layer interm ixing and both the outdiffused In and the underlying Ti become oxidized. The sheet resistivity of the ITO was in the range 10-3-10-4 -cm for all temperatures up to 350Â°C. This is less than the resistivity of other oxides such as NiO (~10-1 -cm) [132,137] wh ich are potential
66 candidates for transparent layers, allowing more uniform carrier injection to the active region in ZnO-based LEDs. 4.4 Summary In conclusion, Ti/Au and Ti/Al/Pt/Au contacts show specific contact resistivity in the range 1.6-2.310-4 -cm2 on lightly n-type ZnCdO laye rs after annealing at 450500ÂºC. The temperature dependence of the speci fic contact resistivit y indicates that the dominant mechanism of current transport is field emission. The creation of oxygen vacancies (Vo) appears to play an impor tant role in the Ohmic nature of the Ti/Au and Ti/Al/Pt/Au contacts through a local increase in electron concentration. The Ti/Au scheme shows superior thermal stab ility compared to Ti/Al/Pt/Au. Ohmic contacts using Ti/Au on Al-doped ZnO had the following properties. Contacts made to heavily Al-doped ZnO with carrier concentrations near 1019 cm-3 show minimum specific contac t resistivity of 6.0 10-8 -cm2 after annealing at 300ÂºC. These contacts are Ohmic as-deposite d, exhibiting very good speci fic contact resistances of 2.4x10-7 -cm2. The dominant mechanism of current transport is tunneli ng, even in the as-deposited contacts, showing the advantage of specifically doping the ZnO. This is a valuable result for applications where lo w temperature processing of the ZnO is important, such as transparent electronics or light emitters on cheap substrates such as glass. ITO/Ti/Al contact stack grown on ZnO pr oduces a contact with a resistance <10-5 -cm2 over a broad range of annealing temp eratures and maintains smooth morphology up to 350Â°C. The ITO refractive index is betw een that of ZnO and air, which reduces the
67 reflection at the ZnO/air inte rface. This approach looks promising as a transparent conducting current spreading layer on ZnO-based LEDs.
68 Table 4-1. Concentration of elements de tected in Ti/Au contacts on ZnCdO/ZnO/GaN (in Atom %). Preparation condition C Au O Ti Zn As received 51.9 45.6 2.5 nd nd Annealing at 450oC 46.8 19.2 26.5 5.0 2.5 Annealing at 500oC 46.3 18.7 26.1 4.1 4.8 Table 4-2. Concentration of elements detected in Ti/Al/Pt/Au contacts on ZnCdO/ZnO/GaN (in Atom %). Preparation condition C Au O Al As received 66.8 31.3 1.9 nd Annealing at 450oC 32.9 15.6 26.3 26.3 Annealing at 500oC 28.7 6.5 29.5 32.3
69 0.4 Âµm Al-Doped ZnO MgO Sapphire Figure 4-1. Cross-sectional TEM image of as-grown Al-doped ZnO on MgO buffer on sapphire substrate.
70 Al2O3 P t ZnCdO ZnO GaN C Figure 4-2. TEM cross-section of ZnCdO la yers grown on a ZnO buffer layer deposited on GaN using sapphire as a s ubstrate. The SADPÂ’s fr om individual layers are also shown.
71 GaN ZnO ZnCdO C Pt Figure 4-3. Higher magnification TEM cross-sec tion images of the as-grown structure.
72 200300400500600 0.0 4.0x1038.0x1031.2x1041.6x104 Ti/Au Ti/Al/Pt/Au Sheet resistance ()Annealing Temperature (oC) Figure 4-4. Sheet resistance as a functi on of annealing temperature for Ti/Auand Ti/Al/Pt/Au contacts on ZnCdO.
73 200300400500600 0 20 40 60 80 Ti/Au Ti/Al/Pt/Au Transfer resistance (-mm)Annealing Temperature (oC) Figure 4-5. Transfer resistance as a functi on of annealing temper ature for Ti/Au and Ti/Al/Pt/Au contacts on ZnCdO.
74 200300400500600 10-410-310-2 Ti/Au Ti/Pt/Al/Au Specific contact resistivity (-cm2)Annealing Temperature (oC) Figure 4-6. Specific contact resistivity as a function of annealing temperature for Ti/Au and Ti/Al/Pt/Au contacts on ZnCdO.
75 Ti n -ZnCdO n+-ZnO Interfacial oxide Figure 4-7. Schematic of pr oposed contact conduction mech anism, through reduction of depletion region by formation of oxyge n vacancies and an interfacial TiOx layer.
76 Reference 600oC 450oC 350oC 10 Âµm Reference 450oC 600oC 350oC Figure 4-8. Optical microscopy images of T i/Au (top) and Ti/Al/Pt/A u contacts (bottom) annealed at differe nt temperatures.
77 0500100015002000 -8000 0 8000 Au O C Au AuAu Au Au AuAs received Count/secKinetic energy (eV) -8000 0 8000 Au Au C Ti Ti O ZnAu Au Au Au AuAnnealing at 450oC -8000 0 8000 Au Au C Ti Ti O Zn AuAu Au Au AuAnnealing at 500oC Figure 4-9. AES surface scans of Ti/Au contacts on ZnCdO/ZnO/GaN as-received (bottom), after annealing at 450oC (middle), and 500oC (top).
78 050010001500200025003000 0 50 100 N Au Ti C O Zn Cd GaAs received Atomic concentration (%) Sputter Depth (Ã…) 0 50 100 C Au Ti O Zn Cd N GaAnnealing at 450oC 0 50 100 C N O Ti Zn Ga Cd AuAu C O Cd Zn Ti N GaAnnealing at 500oC Figure 4-10. AES depth profiles on Ti/Au contacts on ZnCdO/ZnO/GaN as-received (bottom), after annealing at 450oC (middle), and 500oC (top).
79 0500100015002000 -8000 0 8000 Au O C Au Au Au Au Au AuAs received Count/secKinetic energy (eV) -8000 0 8000 Al Au Au C O Au Au Au Au AuAnnealing at 450oC -8000 0 8000 Al Au Au C O AuAuAu Au AuAnnealing at 500oC Figure 4-11. AES surface scans of Ti/Al/Pt/ Au contacts on ZnCdO/ZnO/GaN as-received (bottom), after annealing at 450oC (middle), and 500oC (top).
80 0100020003000400050006000 0 50 100 C N O Al Ti Zn Ga Cd Pt AuN Au Al C Pt Ti OAs received Atomic concentration (%) Sputter Depth (Ã… ) 0 50 100 Au C Al Pt Au Cd Zn N GaAnnealing at 450oC 0 50 100 C Ga O Zn Al Pt Au O Ti Zn Ti N GaAnnealing at 500oC Figure 4-12. AES depth profiles on Ti/Pt/A l/Au contacts on ZnCdO/ZnO/GaN asreceived (bottom), after annealing at 450oC (middle), and 500oC (top).
81 100200300400500 90 100 110 120 130 140 150 160 170 As deposited N ~ 1.32x1019 cm-3 N ~ 9.09x1018 cm-3 Sheet resistance (/)Annealing Temperature (oC)As deposited 100200300400500 0.03 0.06 0.09 0.12 As deposited As deposited N ~ 1.32x1019 cm-3 N ~ 9.09x1018 cm-3 Transfer resistance (-mm)Annealing Temperature (oC) Figure 4-13. Electrical propertie s of Ti/Au contacts on ZnO:Al . (a) Sheet resistance as a function of annealing temperature. (b) Transfer resistance as a function of annealing temperature.
82 100200300400500 1E-7 1E-6 As deposited As deposited N ~ 1.32x1019 cm-3 N ~ 9.09x1018 cm-3 Specific contact resistivity (-cm2)Annealing Temperature (oC) Figure 4-14. Specific contact resistivity as a function of annealing temperature for Ti/Au contacts on ZnO:Al.
83 As deposited Annealing at 150oC Annealing at 250oC Annealing at 450oC Figure 4-15. Optical microscopy images of Ti/Au contacts annealed at different temperatures on ZnO:Al (N ~ 9x1018 cm-3). Figure 4-16. Topographic images of (a) as received Ti/Au deposited on ZnO:Al(N ~ 9.x1018 cm-3.) and (b) after a subsequent anneal to 250 oC. The scan size of these AFM images is 5 Âµm x 5 Âµm.
84 -8000 -4000 0 4000 0500100015002000 -8000 -4000 0 4000 Annealing at 250oC Zn Au O Au Au Zn Au Au C Count/secKinetic energy (eV) Au C O Au Au Au Au Au As received 0 25 50 75 100 125 O Zn Ti C Annealing at 250oC Atomic Concentration (%)Sputter Depth (Ã…) Au050010001500 0 25 50 75 100 125 Zn O Au Ti C C O Ti Zn AuAs received Figure 4-17. AES surface scans (top) and depth profiles (bottom) of Ti/Au contacts on ZnO:Al/sapphire as-received and after annealing at 250oC.
85 100200300400500 80 90 100 110 120 130 1E-7 1E-6 1E-5 1E-4 1E-3 Specific contact resistance (-cm2) As deposited As deposited N ~ 1.33x1019 cm-3 Sheet resistance (/)Temperature (oC) Figure 4-18. Sheet resistance (s quare data points) and specifi c contact resistivity (circle data points) as a function of annealin g temperature for ITO/Ti/Au contacts on ZnO:Al.
86 As received Annealing at 250oC Annealing at 350oC Annealing at 450oC Figure 4-19. Optical microscopy images of IT O/Ti/Au contacts annealed at different temperatures on ZnO:Al (N ~1.3x1019 cm-3.) The inner contact diameter is ~300 Âµ m.
87 0600120018002400 -1x1040 1x104 O C Au Au AuAs received Count/secKinetic energy (eV) -1x1040 1x104 In C In O Au AuAnnealing at 350oC -1x1040 1x104 In C In O ZnAnnealing at 450oC Figure 4-20. AES surface scans of ITO/T i/Au contacts on Al-doped ZnO as-received (bottom), after annealing at 350oC (middle), and 450oC (top).
88 050010001500 0 50 100 C O Ti Zn In AuAu C Ti OAs received Atomic concentration (%) Sputter Depth (Ã…) 0 50 100 C Au In Zn InAnnealing at 350oC 0 50 100 In C O Zn Au O Ti Zn TiAnnealing at 450oC Figure 4-21. AES depth profiles on ITO/Ti/A u contacts on Al-doped n-ZnO as-received (bottom), after annealing at 350oC (middle), and 450oC (top).
89 CHAPTER 5 BANDGAP ENGINEERING 5.1 Introduction InGaN materials system is already comme rcialized over much of the blue/green regions of the spectrum. Th e ZnMgO/ZnCdO system has some advantages because of the high exciton binding energy of ZnO relative to GaN, the availability of high quality ZnO substrates (enabling the fabrication of ve rtical geometry devices with low threading dislocation densities) and the simpler pro cessing relative to GaN for which convenient wet etches are not available [19,21,109]. Ther e have been a number of recent reports on ZnO-based light emitting structures, including homojunctions [83,91,92], hybrid structures involving heterost ructures with SiC, GaN/sa pphire, AlGaN/ sapphire or Cu2O [35,93-97], and ZnO pin homoj unction diodes grown on ScMgAlO4 substrates [15,16]. Ternary ZnCdO seems to be an appropriate candidate for the narrow bandgap active region because of the smaller bandgap of Cd O (2.3 eV) [65,138,139]. In designing LED structures with this system, there is a need to have available basic information such as the valence and conduction band offsets. To date, little is known for the ZnCdO/ZnO system, although the bandgap energies have been reported as a function of Cd composition [65,138]. AlGaN/GaN high electron mobility transistors (HEMTs) are of great interest for high-frequency/high-power/hightemperature applications due to the very high density of the 2D gas and an extremely high breakdown voltage [140-157]. Applications for these devices include high frequency wireless base stations and br oad-band links, commercial
90 and military radar, and satellite communicat ions. The use of metal-oxide-semiconductor (MOS) or metal-insulator-semiconductor (MIS ) gates for HEMTs produces a number of advantages over the more conventional Schot tky metal gates, incl uding lower leakage current and greater voltage swing [149-152] . One of the continuing issues with microwave power AlGaN/GaN HEMTs is su rface and bulk carrier trapping phenomena causing the power performance to degrade s ubstantially at high frequencies and high signal levels. This phenomenon can also be observed as a current dispersion between dc and pulsed test conditions or a degraded rf output power. One technique to mitigate the surface carrier traps is to passivate the HEMT structure with a dielectric layer such as SiNX, Sc2O3, or MgO over the source/gate and ga te/drain regions [141,145,151-157]. In particular, MgO has been shown to provide the most effective reduction of current collapse and also a low interface state densit y with GaN [154-157]. However, to date there is no information available on the band offsets in the MgO/GaN system. This is important for determining the suitability of this dielectric for device operation at elevated temperatures where carrier confinement must still be maintained. The use of gate insulators to reali ze metal-oxide-semiconductor field effect transistors (MOSFETs) for GaN-based tran sistors would provide many advantages relative to the commonly used Schot tky-gate devices [120,147,149-151,154,155,158166]. These include lower leakage current a nd greater voltage swing, improved thermal stability against reaction in the gate regi on, simplification of circuit design since enhancement-mode MOSFETs can be used to fo rm single supply volta ge control circuits for power transistors and reduced power c onsumption through the use of complementary devices. Sc2O3 films deposited by rf plasma-enhanced MBE have been shown to provide
91 low interface state densities (in the 1011 eV-1Â·cm -2 range) on nand p-GaN [154,155,165,166] and also are effective in reduc ing the effect of surface states on current collapse in AlGaN/GaN HEMTs. The Sc2O3 has a bixbyite crystal st ructure, with a 9.2% lattice mismatch to GaN, a high dielectric constant (14) and reasonable bandgap (6.0 eV). Applications for these devices include high frequency wirele ss base stations and broadband links, commercial and military radar, a nd satellite communications. In all cases a complete understanding of the interface is needed but as yet there are no published reports on the band offsets in the Sc2O3/GaN heterostructure. We have determined in the related MgO/GaN system that the bandgap di fference of ~4.36 eV between the MgO and GaN has an almost 3:1 ratio between EC and EV . This is promising for maintaining carrier confinement in n-type devices for high temperature operation. In this chapter, we present an X PS study of the valence band offset ( EV) in a Zn0.95Cd 0.05O/ZnO (0001) heterojunction. The sa mples were grown by MBE and exhibit no detectable phase separation at this concentration. The valence band offset is determined to be Ev =0.17+/-0.03 eV. We also invest igate an XPS study of the valence band offset ( EV) in a MgO/GaN (0001) heterojunction. The MgO layers were grown by MBE on GaN/sapphire templates, with a va lence band offset determined to be Ev =1.06+/-0.15 eV. In addition to this, we also report an XPS study of the valence band offset ( EV) in a Sc2O3/GaN (0001) heterojunction in which the Sc2O3 layers were grown by MBE on GaN/sapphire templates. The vale nce band offset is determined to be Ev = 0.42+/-0.07 eV, which indicates the conducti on band offset is approximately 5 times larger.
92 5.2 Experimental 5.2.1 ZnCdO/ZnO Heterojunction Band Offset MBE growth of ZnO and ZnCdO single layers was performed on c-plane sapphire substrates. Low-temperature K nudsen cells were used to evaporate 6N-purity Zn, Cd and Mg. Atomic oxygen fluxes were produced by a radio-frequency (RF) plasma source. The source was operated at an O2 flow rate of 2-5 standard cubic centimeters per minute, depending on the Zn flux, and a forward power of 150W. In-situ RHEED was used to monitor MBE growth of ZnO and Zn0.95Cd0.05O. After thermal cleaning of the sapphire surface, the substrate temper ature was lowered to 400-600 C. The ZnO growth was initiated by simultaneously opening the Zn and the atomic oxygen shutters to ensure Znpolarity for as-grown films. To reduce th e number of variable growth parameters, TS (400Â°C) and Zn/O flux ratio (1.5:1) were kept constant and only the Cd cell temperature was used to control the com position during growth. After gr owth, RT CL measurements were performed on the samples to define th e spectral position of the near-band-edge emission peaks for ZnCdO. Three samples were used to characterize the ZnCdO/ZnO system. Namely, a 5000Ã… thick ZnO laye r grown on sapphire, 0.1 Âµm ZnCdO/0.1 Âµm ZnO grown on 2 Âµm of GaN on sapphire and 1 nm ZnCdO/0.1 Âµm ZnO grown on 2 Âµm of GaN on sapphire. The experimental bandg ap of our ZnCdO was ~ 2.9 eV from CL measurements at 25Â°C. There was no phase separation in the CdZnO films, as confirmed by CL emission mapping. The surfaces of the specimens were exam ined initially by low-resolution survey scans to determine which elements were pr esent. Very-high-resolution spectra were acquired to determine the binding energy (i.e ., chemical state) and concentration of specific elements observed in the survey spectr a. The quantification of the elements was
93 accomplished by using the atomic sensitivity factors for a Physical Electronics Model 5700LSci XPS spectrometer. The approximate escape depth (3 sin) of the carbon electrons was 80Ã…. A Physical Electronics 5700LSci with Monochromatic aluminum Xray source (energy 1486.6 eV) was used in these experiments. The source power was 350 watts and the analysis region was 2 mm x 0.8 mm. The exit angle was 65Â°. Charge correction was performed by using the known pos ition of the C-(C,H) li ne in the C 1s spectra at 284.8 eV. The electron pass en ergy was 187.85 eV and the entrance slit width was 1.3 mm. The total energy resolution was 0.48+/0.01 eV. Charge neutralization was performed with an electron flood gun. A core-level photoemission-based method was used to determine the valence band offset [51,168]. Appr opriate shallow core-level peaks were referenced to the top of the valence band for the thic k ZnO (which was also checked by comparison with a true bulk ZnO substrate) and the thick film of ZnCdO, using a linear extrapolation method to determine the valence band maximum. The resulting binding energy differences between the co re peaks and valence band minimum for the single layers were then combined with core-level binding energy differences for the heterojunction sample to obtain the valence ba nd offset. This is a standard method for determining band offsets [44,48,58,62, 169]. The XPS was calibrated using a polycrystalline Au foil. The Au f7/2 peak position and Fermi-edge inflection point were determined to be 84.00Â±0.02 and 0.00Â±0.02 eV, respectively. Therefore, all of the binding energies (BE) are accurate on an absolute scale within 0.02Â–0.03 eV, over the binding energy range of 0 to 100 eV.
94 5.2.2 MgO/GaN Heterojunction Band Offset The MgO growth was performed in a m odified RIBER 2300 MBE equipped with a reflection high-energy electr on diffraction (RHEED) system. Oxide growth was performed using a standard effusions oven containing Mg (99.99%) operating at 360Â°C. Atomic oxygen was supplied from an Oxford MPD21 radio frequency plasma source with 300 watts forward power at 8x10-6 Torr oxygen pressure. The substrate temperature was measured using a backside thermocouple that was calibrated using the melting points of In (156Â°C), InSb (525Â°C) and GaSb (712Â°C) . Three samples were used to characterize the MgO/GaN system. Namely, a 3 Âµm thick GaN layer grown by Metal Organic Chemical Vapor Deposition on a c-plane sa pphire substrate, 0.04 Âµm MgO grown on 3 Âµm of GaN on sapphire and 5 nm MgO/ 3 Âµm of GaN on sapphire. The crystal structure of magnesium oxide is rocksalt, which is an FCC array of magnesium atoms with an interpenetrating FCC array of oxygen atoms. The (111)//(0001) is the lowest energy configuration for the interface of these two ma terials and the expected growth plane for the MgO, which produces the lattice mism atch between the MgO(111) and the GaN (0001) of -6.5%. At a substrate temperature of 100Â°C, the single crystal nature of the oxide film appears to remain for the duration of the growth and the growth rate is 2.5-3.0 nm/min. From the cross-section TEM micr ographs, the MgO becomes polycrystalline after 4nm of growth. However, the film act ually rotates, mainta ining the (111)//(0001) symmetry as no other XRD peaks were seen other than the (222). The surfaces of the specimens were exam ined initially by low-resolution survey scans to determine which elements were pr esent. Very-high-resolution spectra were acquired to determine the binding energy (i.e ., chemical state) and concentration of specific elements observed in the survey spectr a. The quantification of the elements was
95 accomplished by using the atomic sensitivity factors for a Physical Electronics Model 5100LSci XPS spectrometer as described in Section 5.2.1. A Physical Electronics 5100LSci with aluminum X-ray source (e nergy 1486.6 eV) was used in these experiments. The source power was 300 wa tts and the analysis region was 10 mm x 4 mm. The exit angle was 45Â°. The electron pass energy was 17.9 eV and the entrance slit width was 1.3 mm. The total energy resolu tion was 0.05+/0.01 eV. A core-level photoemission-based method was used to dete rmine the valence band offset [51,168]. Appropriate shallow core-level p eaks were referenced to the top of the valence band for the GaN and the thick film of MgO , using a linear extrapolation method to determine the valence band maximum. The resulting binding energy differences between the core peaks and valence band minimum for the single layers were then combined with corelevel binding energy differences for the heterojunction sample to obtain the valence band offset [44,51,168]. This is a standa rd method for determining band offsets [48,58,62,64,169]. 5.2.3 Sc2O3/GaN Heterojunction Band Offset Scandium oxide was deposited epitaxia lly on (0001) GaN in a MBE using elemental Sc and atomic oxygen supplied from a Oxford MPD21 radio frequency plasma source with 300 watts forward power at 810-6 Torr oxygen pressure [157,158]. All oxide growths were performed in a m odified RIBER 2300 MBE equipped with a reflection high-energy electron diffraction (RHEED) system. A standard effusion cell operating at 1190Â°C was used for the evapor ation of the scandium (99.999%). The substrate temperature was measured using a backside thermocouple that was calibrated using the melting points of In (156Â°C), In Sb (525Â°C) and GaSb (712Â°C). Previous
96 measurements have found these growth condi tions produce an interface trap density of 51011 eV-1Â·cm-2 at Ec-Et = 0.2eV determined from the Terman method and 1.111012 eV1Â·cm-2 at Ec-Et = 0.42eV from AC conductance measurements conducted at 300C [154,155,165]. Three samples were used in our XPS experiments, namely, a 3 Âµm thick GaN layer grown by Metal Organic Chemical Vapor Deposition on a c-plane sapphire substrate, 0.04 Âµm Sc2O3 grown on 3 Âµm of GaN on sapphire and 3 nm Sc2O3/ 3 Âµm of GaN on sapphire. The crystal structure of s candium oxide is Bixbyite, which is an FCC array of scandium atoms with oxygen occupyi ng Â¾ of the tetrahedral sites. The symmetry of the (111) of the FCC array is iden tical to the (0001) of the hexagonal array. The (111)//(0001) is the lowest energy c onfiguration for the interface of these two materials and the expected growth plane for the Sc2O3, which allows for the lattice mismatch between the Sc2O3 (111) and the GaN (0001) to be 9%. When grown at a substrate temperature of 100Â°C, the single crystal nature of th e oxide film is lost after a few nanometers and the remaining growth is polycrystalline. The surfaces of the specimens were exam ined initially by low-resolution survey scans to determine which elements were pr esent. Very-high-resolution spectra were acquired to determine the binding energy (i.e ., chemical state) and concentration of specific elements observed in the survey spectra . The qua ntification of the elements was accomplished by using the atomic sensitivity factors for a Physical Electronics Model Perkin-Elmer PHI 5100 XPS spectrometer as described in Section 5.2.1. A Physical Electronics Perkin-E lmer PHI 5100 with magnesium X-ray source (energy 1253.6 eV) was used in these experi ments. The source power was 300 watts and the analysis region was 10 mm x 4 mm. The exit angle was 45Â°. Charge correction was
97 performed by using the known position of the C(C, H) line in the C 1s spectra at 284.5 eV. The electron pass energy was 17.9 eV and the entrance slit width was 1.3 mm. The total energy resolution wa s 0.05+/0.01 eV. Core level survey spectra of the 0.04 Âµm thick Sc2O3, 3 nm layer of Sc2O3 on GaN/sapphire, and the GaN/sapphire templates using a pass energy of 89.45 eV at takeoff angle of 45o are shown in Figure 5-1. A core -level photoemission-based method was used to determine the valence band offset [51,157,168]. Appropriate shallow core-level peaks were referenced to the top of the valence band for the GaN and the thick film of Sc2O3, using a linear extrapolation method to determine the valence band maximum. The resulting binding energy differences between th e core-level peaks and valence band minimum for the single layers were then combin ed with core-level binding energy differences for the heterojunction samp le to obtain the valence band offset [51,58,62,64,157,168,169]. 5.3 Results and Discussion 5.3.1 ZnCdO/ZnO Heterojunction Band Offset Figure 5-2 shows the XPS Zn 2p3 narrow scan and valence band spectrum from the 0.1 Âµm ZnCdO/0.1 ÂµmZnO/MOCVD GaN/C-pl ane sapphire and ZnO substrate samples using a pass energy of 11.75 eV and step size of 0.025 eV. The valence band value (EV) was determined by linearly fitting the leading edge of the valence band and linearly fitting the flat energy distribution and finding the inte rsection of these two lines, as shown in the insets of the Figure 5-2. Core level survey spectra of ZnCdO, 1nm layer of ZnCdO on ZnO, and a ZnO substrate using a pass en ergy of 187.85 eV at take-off angle of 65o are shown in Figure 5-3. Table 5-1 shows a summary of the band offset results. These values were then inserted into the following equations to calculate Ev, namely Ev =
98 (EZn-2p-EV)ZnO-( EZn-2p-EV)thick ZnCdO(EZn-2p-EZn-2p)ZnCdO/ZnO. The resulting Ev was 0.17+/-0.03 eV for the Zn0.95Cd0.05O/ZnO (0001) heterojunction. This is relatively good value for device applications in which strong ca rrier confinement is needed, such as light emitters or heterostructure field effect transistors. For example, the valence band offsets in the In0.2Ga0.8N/GaN system are of order 0.06 eV . Figure 5-4 shows a schematic of the energy band lineup in the ZnCdO/ZnO heterostructure, with all of the energy scales include d. The bandgap of the ZnO used here is 3.37 eV at room temperature, as determined by photoluminescence measurements. The bandgap difference of 0.47 eV between the Zn0.9Cd 0.1O and ZnO has an almost 2:1 ratio between EC and EV. Similar work is needed in the ZnMgO/ZnO system. There has been some initial recent work on band alignment in CdS/ZnxMg1-xO interfaces suggesting the valence band offset between ZnO and ZnMgO should be small , but other data show conflic ting results [98,171]. 5.3.2 MgO/GaN Heterojunction Band Offset Figure 5-5 shows the XPS Ga 3d narrow scan and valence band spectrum from the 0.04 Âµm MgO/MOCVD GaN/C-pl ane sapphire and GaN/sapphire template samples using a pass energy of 17.9 eV and step size of 0.05 eV. The valence band value (EV) was determined by linearly fitting the leading edge of the valence ba nd and linearly fitting the flat energy distribution and finding the inte rsection of these two lines, as shown in the insets of the Figure 5-5. Core level survey spectra of MgO, 5nm layer of MgO on GaN/sapphire, and the GaN/sapphire templates using a pass energy of 44.75 eV at takeoff angle of 45o are shown in Figure 5-6. Table 5-2 shows a summary of the band offset results. These values were then inserted into the following equations to calculate Ev, namely Ev = (EGa-3d-EV)GaN-( EMg-2p-EV)thick MgO(EGa-3d-EMg-2p)MgO/GaN. The resulting
99 Ev was 1.06+/-0.15 eV for the MgO/GaN (0001) he terojunction. This is an excellent value for high temperature device applications in which strong carrier confinement is needed. For example, a simple calcula tion suggests that at least 0.8 eV band discontinuity is desirable for device ope ration at temperatures up to 500Â°C. Figure 5-7 shows a schematic of the energy band lineup in the MgO/GaN heterostructure, with all of the energy scales include d. The bandgap of the GaN used here is 3.44 eV at room temperature, as determined by photoluminescence measurements. The bandgap difference of ~4.36 eV between the MgO and GaN has an almost 3:1 ratio between EC and EV. Similar work is needed for ot her promising oxide dielectrics for GaN, including Sc2O3 and MgCaO [154,172]. MgCaO can be produced that is lattice matched to the GaN. Recently we have found that HEMTs passivated with Mg0.5Ca0.5O and Mg0.25Ca0.75O showed higher passivation effectiven ess (90% of dc current) than the MgO passivated HEMTs (83% dc current). This is due to the closer lattice matching of these calcium containing oxides and the redu ction in interface traps associated with lattice mismatch . 5.3.3 Sc2O3/GaN Heterojunction Band Offset Figure 5-8 shows the XPS Ga 3d narrow scan and valence band spectrum from the 0.04 Âµm Sc2O3/MOCVD GaN/C-plane sa pphire and GaN/sapphire template samples using a pass energy of 17.9 eV a nd step size of 0.05 eV. The valence band value (EV) was determined by linearly fitting the leading edge of the valence band and linearly fitting the flat energy distribution and finding the inte rsection of these two lines, as shown in the insets of the Figure 5-8. Table 5-3 shows a summary of the band offset results. These values were then inserted into the following equations to calculate Ev, namely Ev = (EGa-3d-EV)GaN-( ESc-3p-EV)thick Sc2O3(EGa-3d-ESc-3p) Sc2O3/GaN. The resulting Ev was
100 0.42+/-0.07 eV for the Sc2O3/GaN (0001) heterojunction. Based on the experimental value of 6 eV for the bandgap of our Sc2O3 films, this translates to a conduction band offset EC of 2.14 eV for the Sc2O3/GaN heterojunction or an almost 5:1 ratio for EC/ EV. For comparison, the valence band offset for the Sc2O3/GaN system is 1.06 eV and the conduction band offset was 3.30eV , wh ich may make the latter a better choice for very high temperatur e device applications. Figure 5-9 shows a schematic of the energy band lineup in the Sc2O3/GaN heterostructure, with all of the energy scales included. The bandgap of the GaN used was 3.44 eV at room temperature, as determined by photoluminescence measurements on our films. 5.4 Summary In summary, XPS determination of the valence band offset of Zn0.95Cd0.05O / ZnO(0001) heterojunctions shows a value of 0.17 eV. Given the bandgap difference of 0.47 eV between the two materials, this tran slates to a nested interface band alignment with a conduction band offset of 0.30 eV. This shows that reasonable valence and conduction band offsets can be obtained in this materials system at Cd concentrations low enough to avoid phase separation. Future wo rk should include similar measurements on the ZnMgO/ZnO and ZnMgO/ZnCdO systems th at are promising for visible and UV LEDs and especially focus on the influence of Mg and Cd composition on the band offsets. Regarding the MgO/GaN band offset study, XPS determination of the valence band offset of MgO /GaN hetero junctions shows a value of 1.06 eV. Given the bandgap difference of ~4.36 eV between the two material s, this translates to a nested interface band alignment with a conduction band offset of ~3.30 eV. This shows that good valence
101 and conduction band offsets can be obtained in th is materials system a nd is well-suited to high temperature applications. Future work should include similar measurements on the MgCaO/GaN and Sc2O3/GaN systems that are also promising for power HEMT gate dielectrics and surface passivation and especially focus on the effect of Ca composition on the band offsets in the former system. Finally, the valence band offset of Sc2O3 /GaN heterojunctions determined by XPS is found to be 0.42 eV. Given the bandgap difference of ~2.56 eV between the two materials, this translates to a nested in terface band alignment with a conduction band offset of ~2.14 eV. This shows that good valence and conduction band offsets can be obtained in this materials system and is we ll-suited to high temper ature applications.
102 Table 5-1. Values of ZnCdO/ZnO band o ffsets determined in these experiments. ZnCdO VBM (Bulk ZnCdO) Zn 2p3 (Bulk ZnCdO) Zn 2p3 ZnCdO VBM ZnO VBM (Bulk ZnO) Zn 2p3 (Bulk ZnO) Zn 2p3 ZnO VBM Zn 2p3-Zn 2p35 (ZnCdO/ZnO) Valence band offset Ev 1.18 1020.85 1019.67 1.33 1020.83 1019.50 0.00 -0.17 Table 5-2. Values of MgO/GaN band offs ets determined in these experiments. GaN VBM Ga 3d Ga3dGaNVBM MgO VBM Mg 2p Mg 2p MgOVBM Ga 3d-Mg 2p Valence band offset (Bulk GaN) (Bulk GaN) (Bulk MgO) (Bulk MgO) (MgO/GaN)Ev 1.78 19.47 17.69 2.13 49. 07 46.94 -30.31 1.06 Eg of MgO: 7.8eVEg of GaN: 3.44eVconduction band offset: 3.30eV. Table 5-3. Values of Sc2O3/GaN band offsets determined in these experiments. GaNVBM Ga 3d Ga 3dGaNVBM Sc2O3 VBMSc 3p Sc 3pSc2O3 VBM Ga 3d -Sc 3p Valence band offset Ev (Bulk GaN) (Bulk GaN) (Bulk Sc2O3) (Bulk Sc2O3) (Sc2O3/GaN) 1.91 19.53 17.62 2.34 30.99 28.65 -11.45 0.42 Eg of Sc2O3: 6 eV, Eg of GaN: 3.44eV, and conduction band offset: 2.14eV.
103 10008006004002000 N 1s Ga 3s Sc2s Ga 2p N 1s Sc2s Ga3dGaNSc 3p Ga LMM Sc 2p Ga 3d Sc 3p Sc2p O 1s O 1s O KLL Sc LVV O KLL Sc LVV O KLL N 1s O 1s Ga LMM Ga3pGa3p Ga3d Ga 2p30 Ã… Sc2O3/GaN 400 Ã… Sc2O3/GaN Intensity (a.u.)Binding Energy (eV) Figure 5-1. Core level survey spectra of Sc2O3, 3 nm layer of Sc2O3 on GaN/sapphire, and a GaN/sapphire template using a pass energy of 89.45 eV at take-off angle of 45o.
104 Figure 5-2. XPS Zn 2p3 narrow scan and valence band spectrum of 0.1 ÂµmZnCdO/0.1 ÂµmZnO/MOCVD GaN/C-plane sap phire and ZnO substrate.
105 Figure 5-3. Core level survey spectra of ZnCdO, 1nm layer of ZnCdO on ZnO, and a ZnO substrate using a pass energy of 187.85eV at take-off angle of 65o.
106 EC ZnCdO EV ZnCdO EC ZnO EV ZnO E C =0.30eV E V =0.17eV ZnCdO ZnO EZn 2 p 3 ZnO (EV Â– EZn 2p3)ZnO=1020.83 eV Eg ZnO =3.37 eV Eg ZnCdO=2.90 eV EZn 2 p 3 ZnCdO (EV Â– EZn 2p3)ZnCdO =1020.85eV Figure 5-4. Energy band diagram of thin ZnCdO/ZnO heterojunction interface. EB is the corresponding core level separa tion measured across the interface.
107 55504550 25201550 543210 1.78 eV VBM 543210 2.13 eVVBM Intensity (a.u.)Binding Energy (eV)400 Ã… MgO/GaNIntensity (a.u.)Binding Energy (eV) EMgO Mg 2p EMgO VMg 2p Ga 3d Intensity (a.u.)Binding Energy (eV)GaNEGaN Ga3d EGaN V Figure 5-5. XPS Ga 3d narro w scan and valence band sp ectrum of 40 nm MgO/MOCVD GaN/C-plane sapphire and GaN/sapphire template.
108 120010008006004002000 N 1s Ga3dGaNMg 2p Mg 2s Mg KLL Mg 2p Mg 2s Mg KLL O 1s O 1s O KLL Mg1s O KLL Mg1s O KLL N 1s O 1s Ga LMM Ga3s Ga3p Ga3d Ga 2p50 Ã… MgO/GaN 400 Ã… MgO/GaN Intensity (a.u.)Binding Energy (eV) Figure 5-6. Core level survey spectra of MgO, 5 nm layer of MgO on GaN/sapphire, and a GaN/sapphire template using a pass en ergy of 44.75 eV at take-off angle of 45o.
109 EV MgO EMg 2p MgO EC GaN EV GaN EC MgO EC=3.33 eV EV=1.03 eV (EV Â– EM g2p)MgO =46.88 eV Eg MgO =7.8 eV Eg GaN=3.44 eV EGa 3d GaN (EV Â– E 3d)GaN =17.73 eV (E Ga 3d Â– E Mg 2p)MgO/GaN =30.31 eV Figure 5-7. Energy band diagram of th in MgO/GaN heteroju nction interface. EB is the corresponding core level separation measured across the interface.
110 35302550 25201550 543210 2.34 eV VBM ESc2O3 Sc 3p ESc2O3 V Intensity (a.u.)Binding Energy (eV)400Ã… Sc 2 O 3 /GaNIntensity (a.u.)Binding Energy (eV)Sc 3p543210 1.91 eV VBM Ga 3d Intensity (a.u.)Binding Energy (eV)GaN EGaN Ga3d EGaN V Figure 5-8. XPS Ga 3d and Sc 3p narrow s cans and valence band spectra of 40 nm Sc2O3/MOCVD GaN/C-plane sapphire a nd GaN/sapphire template.
111 EC GaN EV GaN EC Sc2O3 EV Sc2O3 EC=2.14 eV EV=0.42 eV GaN Sc2O3 ESc 3p Sc2O3 (EV Â– ESc 3p)MgO =28.65 eV Eg Sc2O3 =6.0 eV Eg GaN=3.44 eV EGa 3d GaN (EV Â– E Ga 3d)GaN =17.62 eV (E Ga 3d Â– E Sc 3p)Sc2O3/GaN =11.45 eV Figure 5-9. Energy band diagram of thin Sc2O3/GaN heterojunction interface.
112 CHAPTER 6 ZINC OXIDE-BASED LIGHT-EMITTING DIODES SIMULATION 6.1 Introduction Light-emitting diodes have a smaller size, longer lifetime, and higher efficiency than the traditional light sources . LED development is focused on achieving increasing output power and efficiency as well as a wider range of color. The ultraviolet (UV) light sources are of high interest for white light generation. A number of recent papers have reported ZnO-based light emitting structures [15,16,35,83,91-97]. However, device performance was limited due to the hi gh spreading resistance of the bottom ZnO layer, suggesting the need for a conducting ZnO substrate. For all these ZnO LED structures, the growth has not yet been optimi zed. Another important factor in the design of ZnO-based LED is the realization of bandga p engineering to create barrier layers and quantum wells using heterostructures. With re spect to higher bandgaps, an increase up to 4.0 eV has been achieved by the incorporat ion of Mg into the ZnO layer while still maintaining the wurtzite structure . Ternary ZnCdO seems to be an appropriate candidate for narrow bandgap applications be cause of the smaller bandgap of CdO (2.3 eV) . Furthermore, when grown on c-plane sapphire, analogous to III-nitride, the total polarization of ZnO is aligned al ong the (0001) growth direction. Thus, polarization-induced fields may influence th e electronic band structure and carrier concentration profiles of ZnO-base LEDs. Although many factors need to be considered in designing a ZnO-based LED, no comprehe nsive investigation on optimized device structures and their expected performance is currently availa ble. Therefore, there is a
113 clear need for providing some design parameters for the LED structures in terms of layer structure, doping and compos ition and how they affect li ght emission intensity and current-voltage (IÂ–V) characteristics. In this chapter, we present physical-based onedimensional (1D) simulation using SiLENSe simulation software. Simulation has been used to characterize the properties of ZnMgO/ZnCdO/ZnMgO/ZnO substrate LEDs. A physical model is used to optimize material parameters to maximize the light intensity from the device. Section 6.2 describes the th eoretical models and the material parameters used. The simulation results are presented in section 6.3. 6.2 Models and Parameters The simulation program used in this st udy was SiLENSeÂ™ . This is a 1D simulator which can model a band diagram, carri er injection and reco mbination, and light emission profiles in wide bandgap LED heterostructures. SiLENSe 2.1 is capable of simulating heterostructures made not only of gr oup-III nitrides, but also of other wurtzite semiconductors (for example, ZnMgO all oys) including hybrid st ructures. The LED operation of the heterostructure is consider ed within the framework of the 1D drift diffusion model of carrier tran sport that accounts for specific features of the nitride semiconductors including a strong piezoeffect , the existence of s pontaneous electric polarization, low efficiency of acceptor activat ion, and high threading dislocation density (normally, ~107Â–109 cm 2). Along with bimolecular radiative electron and hole recombination, an original model of nonradiative carrier recombination at threading dislocation cores is also cons idered. The latter allows an analysis of the interaction between the radiative an d nonradiative recombination channe ls and predicts the internal emission efficiency of the LED structure as a function of threading dislocation density. The spectrum of light emission from a singleor multiple-quantum-well active region can
114 be calculated to account for the comple x valence band structure of wurtzite semiconductor by using the 8x8 Kane Hamiltoni an. Energies and wave functions of localized carrier states are found from a nu merical solution of the SchrÃ¶dinger equation within the effective-mass approximation. Gene ration of the grid for each quantum well is totally automated. The model implemented into the code in corporates the following: (i) Localized and distributed polarization charges in th e LED structure induced by both spontaneous and piezopolarization in nitride semiconductors ; (ii) Fermi statistic s for electrons and holes for both degenerate and nondegenerate semiconductors; (iii) pa rtial ionization of donors and acceptors depends on the respective quasi-Fermi level positions;(iv) strain calculations in the LED structure assume coherent growth of all epilayers on an underlying buffer layer; (v) bimolecular ra diative electron and hole recombination neglects quantum-confined effects on the re combination rate; a nd (vi) nonradiative carrier recombination in the principal cha nnel and on threading di slocation cores. The LED IÂ–V characteristics are calculated by th e software at a given serial resistance that is assumed to account for both the lateral current spreading in the LED chip and Ohmic contact resistances. Moreove r, the light emission sp ectra are determined with a post-processing module that uses the ca lculated band profiles of the LED structure and takes into account the complex structure of the valence band of nitride materials and the contribution of the confin ed electronic states. In our simulations we used the available ZnMgCdO material parameters [175, 176], but not all the required parameters are available. For example, the band o ffsets between ZnMgO/ZnO and ZnCdO/ZnO are not known. Whenever the required paramete r was unavailable we used preliminary
115 values obtained from x-ray photoelectr on spectroscopy for samples grown by MBE on cplane sapphire. Given the preliminary nature of the available data, the simulated results should be used as a guide for identifying im portant parameters in the LED design rather than obtaining device parameter values. Th e following describes the details of the theoretical models and the material parameters used. 6.2.1 Strain and Piezoeffect CdyMgxZn1-x-yO is typically grown along the c axis of the wurtzite crystal, which is parallel to the z axis in our coordinate system. The substrate or buffer layer with an inplane lattice constant is shown in Figure 6-1 . The in-pla ne lattice constant aE that follows the Vegard law, and varies with the material composition is expressed by Equation 6-1. aayaxaxyECdOMgOZnO () 1 (6-1) where aCdO,aMgO, and aZnO are the lattice constant for th e respective binary compound, x and y are the molar fractions of MgO and CdO in the CdyMgxZn1-x-yO alloy. The lattice mismatch between the epilayer and the substr ate is defined in Equation 6-2. ()[()]/()ZazaaZEsE (6-2) The thickness of the substrate here is assumed much larger than the total thickness of the LED heterostructure. The strain of the epilay er is considered to be uniform due to the negligible bending of the structure. The Z electronic polarization is given by Equation 63. Only the Z component was considered in these calculations because CdyMgxZn1-x-yO is typically grown along the c axis of the wurtzite crystal. The electric polarization, which is only considered z-component used here, is given by PPeeCCzz s 0 313313332 (/) (6-3)
116 where Pz s is the spontaneous polarization vector, eij is the components of the piezoelectric tensor, and Cij is the elastic stiffness te nsor given in the Voight notation. 6.2.2 Carrier Concentration The electrons and holes in a degenerate semiconductor or electrons in a metal follow the Fermi-Dirac distribution functi on [12,174]. The Fermi-Dirac distribution function is used to determine the electron densities in a metal or in a heavily doped semiconductor. The carrier concentrati on, either n or p-type, is given by nNF FEq kT pNF EFq kTc nc v vp 1212 //(),() (6-4) where F1/2 is the Fermi-Dirac integral, q is the el ectron charge, k is BoltzmannÂ’s constant, T is the absolute temperature, Nc is the effective density stats in the conduction band, Nv is the effective density states in the valence band and is the electric potential. The density of states indicates th e number of states that can be occupied by electrons. The effective density of states is expressed in Equation 6-5. N mkT N mkTc n av v p av 2 2 2 22 32 2 32(),()// (6-5) where is the Plank constant, mn av and mp av are the averaged electron and hole effective masses. When the impurity atoms are introduced , the Fermi level should adjust itself to preserve charge neutrality [ 33]. The number of ionized dono rs and acceptors is given by N N g FEEq kT N N g EEFq k T D D D ncD A A A vAp 1 1 exp() , exp() (6-6) where gD=2 and gA =4 are the degeneracy factors, and ED and EA are the activation energies of electrons and holes. gD equals 2 because a donor le vel can accept one electron with either spin or can have no electron. gA equals 4 because each acceptor impurity level
117 can accept one hole of either spin and the impur ity level is doubly degenerate as a result of the two degenerate valence bands . 6.2.3 Radiative and Non-radiative Recombination In light-emitting diodes, th e electrons and holes in the semiconductor recombine either radiatively or non-radiatively . Radiative recombination is preferred, and produces photon emission. Electrons and holes should exist at the same location and at the same time. To improve the LEDs, it is important to maximize the radiative processes and reduce the non-radiative recombination. The most common non-radiative recombina tion comes from the defects in the material crystals. Any semiconductor material always has some native defects. Even the purest semiconductor has the impurities of ~1012 cm-3. Other sources of non-radiative recombination includes Schockley-Read, Auger, and surface recombination. The total recombination rate is shown in Equation 6-7. R=Rdis+Rrad (6-7) where Rdis is the non-radiative r ecombination rate, and Rrad is the radiative recombination rate. The recombination rate is strongly dependent of the material properties that is described more detail below. In double hetero strucutres, the defects will occur at the interface of the two semiconducto rs if differences are found in the lattice constant and crystal structures between the two materials. Since the high threading dislocation density is the primary cause for the non-radiative reco mbination for nitride epitaxial structures, the non-radiateive recombination rate used in SiLENSe is based on the Shockley-Read approach, which is given by equation 6-8. R npnp nnpp nn EF kT pp FE kTdis dd pdnd d dn d pd ()() ,exp(),exp() (6-8)
118 where n is the total electron concentrati on, p is the total hole concentration, Fn is the electron quasi-Fermi level, Fp is the hole quasi-Fermi level, and Ed is the energy level associated with dislocation traps. The lifet ime of electrons and holes are expressed in Equation 6-9 . np npdd np npDNaN D aVS, , , ,ln() 1 4 13 2 22 (6-9) where Dn,p is the diffusion coefficient of electrons and holes, Nd is the dislocation density in the material, a is the in-plane lattice consta nt (radius of a disloca tion core), and S is the fraction of the electrically active sites on the surface of a dislocation core. Using the bimolecular equation, th e recombination rate is given by RBnp FF kTrad np 1exp (6-10) where B is the temperature-dependent recomb ination constant, n is the total electron concentration, p is the total hole concentration, Fn is the electron quasi-Fermi level, and Fp is the hole quasi-Fermi level. 6.2.4 Light Emission Efficiency Ideally, LEDs should have a quantum effici ency of unity if the active region of LEDs emits one photon for every electron inje cted. In LEDs, the internal emission efficiency is defined by jjrad/ (6-11) where j is the toal current and jrad corresponds to the current de nsity totally converted into the emitted light. The carrier injection effi ciency may decrease due to the non-radiative recombination.
119 6.3 Results and Discussion Figure 6-2 shows a schematic of the ZnMgO/ZnCdO/ZnMgO structure on a ZnO substrate and a corresponding band diagram, as generated by the simulation. The bias condition in the band diagram was 0 V. The range of the various parameters for the simulations is summarized in Table 6-1. The triangular wells arising from the polarization contribution at the interfaces were observed. The magnitudes are much smaller than those for the hybrid ZnO/GaN st ructure. Our preliminary results have yielded small band offsets in the ZnMgO/Zn O system (~0.14 eV in the conduction band and 0.02 eV in the valence band at ~5 at.% Mg). Thus, it is expected that the use of Cd to reduce the bandgap should increase these offsets. Simulated emission spectra were obtained from MgZnO/ZnCdO/ MgZnO as a function of active layer parameters are show n in Figure 6-3. Light output intensity increased with both active layer thickness a nd Cd composition, as well as the expected red shift of the wavelength with increasi ng Cd content, over the range studied. Alternatively, the dependence of doping con centration on emission intensity was not as significant when compared to layer thickness. The series resistances of the active layer had no observable effect on the I Â– V characteristics. This indicates that the LED optical properties are very sensitive to the physical parameters of the active region, and the optimization of the active layer is more criti cal to the emission properties. The simulated emission wavelength was dependent on the ac tive layer composition and the choice of bowing parameters, as expected. Figure 64 shows both experimental data for the emission wavelength of ZnCdO as a func tion of composition, as determined by Rutherford backscattering measurements. The peak emission was obtained from the simulated spectral data for double heterostruct ures, while varying the concentration of
120 Cd. Since there is no phase separation in th e ZnCdO films, as confirmed by CL emission mapping, this fit is a practical calibration for adjustment of bowing parameters for ZnCdO bandgap dependence. In our case we use a bowing parameter of around 6 for a Cd composition of 5%. When considering ZnMgO as the cladding layer, as the thickness increased the I Â– V characteristics showed a gr eater series resistance, but had a low influence on the emission intensity. The most important variab le influencing the IV characteristics was the Mg concentration, see Figure 6-5 (a), be cause of the increased series resistance at high Mg content which also affected the emissi on intensity as shown in Figure 6-5 (b). The higher the Mg composition, the lower th e emission intensity. The doping of the cladding layer also affected the I Â– V characteristics [Figure 6-5 (c)], but had only a minor impact on emission intensity. 6.4 Summary Our work identified several key aspects of ZnO-based LEDs design, focusing on ZnMgO/ZnCdO heterojunctions gr own on ZnO substrates. It is still necessary to obtain accurate measurements of the band offsets as a function of Mg and Cd composition. Active layer thickness and doping densities ar e important factors effecting emission intensity in these structures. The emission intensity increases as the active layer and thickness increase.
121 Table 6-1. Simulated parameters conduc ted in ZnMgO/ZnCdO/ZnMgO. Bolded conditions in the table correspond to th e referenced values in Figure 6-2. Parameters Conditions Active layer thickness 30, 50, 75, 100, 150, 200 (nm) Active layer doping Undoped, 1015, 1016, 1017, 1018, 1019 (cm-3) Active layer composition Cd composition: 0.03, 0.05, 0.07 Cladding layer thickness 300, 400, 500 (nm) Cladding layer doping 1016, 1017, 1018, 1019 (cm-3) Cladding layer composition Mg composition: 0.05, 0.1, 0.15, 0.2, 0.3
122 0 Z d aEas 0 Z d aEas Figure 6-1. Epitaxial film on a substrate.
123 n-Zn0.9Mg0.1O (400nm) p-Zn0.9Mg0.1O (400nm) ZnOsubstrate Zn0.95Cd0.05O(100nm) n-Zn0.9Mg0.1O (400nm) p-Zn0.9Mg0.1O (400nm) ZnOsubstrate Zn0.95Cd0.05O(100nm) 0400800 -6 -4 -2 0 Energy (eV)Distance (nm) ECEVEF@ 0V Active region 0400800 -6 -4 -2 0 Energy (eV)Distance (nm) ECEVEF@ 0V Active region Figure 6-2. Schematic 2D view of ZnO-base d LED structure (top) and its band diagram (bottom).
124 350360370380390 Intensity (a. u.)Wavelength (nm) 100nm 30nm 50nm 75nm 150nm 200nm 350360370380390 Intensity (a. u.)Wavelength (nm) 100nm 30nm 50nm 75nm 150nm 200nm 350360370380390 Intensity (a. u.)Wavelength (nm) Cd:0.05 Cd:0.03 Cd:0.07 350360370380390 Intensity (a. u.)Wavelength (nm) Cd:0.05 Cd:0.03 Cd:0.07 Figure 6-3. Simulated emission spectra fr om ZnMgO/ZnCdO/ZnMgO structure as a function of both active layer thickness (top) and Cd composition (bottom).
125 00.050.100.150.200.250.30 1.5 2.0 2.5 3.0 CL-emission energy (eV)Cd-mole Fraction7 6 5 4.5 8 00.050.100.150.200.250.30 1.5 2.0 2.5 3.0 CL-emission energy (eV)Cd-mole Fraction7 6 5 4.5 8 Figure 6-4. Experimentally observed change of room-temperature cathodoluminescence emission energy (solid circle) and simu lated bowing parameters (dashed line) as a function of Cd mole fraction.
126 Figure 6-5. Simulation of LED behavior for ZnMgO/ZnCdO/ZnMgO structures. (a) I-V characteristics as a function of clad layer composition. (b) Emission spectrum as a function of clad layer composition. (c) I-V characteristics as a function of clad layer doping.
127 CHAPTER 7 CONCLUSIONS AND FUTURE WORK To design and fabricate a ZnO-based li ght-emitting diode (LED), detailed information is needed regarding wet etchi ng, metal semiconductor contacts, and valence and conduction band offsets. Thes e results are presented here. Wet etching was used because it has a high degree of selectivity, is more cost effective, and causes less surface damage than other techniques. In this study, HCl and H3PO4 were used to etch ZnCdO and ZnMgO. In order to control th e etching rate on the order of 10 to 100 nm/min, the solutions were diluted to 10-2 M. First, the etching mechanism was found to be diffusion limited. The selectivity of HCl solution on both ZnCdO/ZnO and ZnMgO/ZnO was higher than that of H3PO4. HCl has a selectivity of 30 to 50 for ZnCdO/ZnO, while the maximum selectivity of H3PO4 is near 18. For the ZnMgO/ZnO system, the selectiv ity of HCl is in the rang e of 300 to 400, whereas the selectivity of H3PO4 is around 50. The activation energy for ZnCdO and ZnMgO was 0.37 and 3.29 kcal/mol, respectively, when etched with HCl. When H3PO4 was used, the activation energies for ZnCdO and ZnMgO we re 0.38 and 2.07 kcal/mol, respectively. The etching rate was improved significantly when the fluid agitation duri ng the etching process was increased. Under these conditions , the etching rate for ZnO was found to be reaction limited and the effects of the etchi ng solution were indepe ndent of agitation. The activation energy for ZnO dissolution was 5.59 kcal/mol for HCl and 5.88 kcal/mol for H3PO4.
128 Metal contacts were deposited on ZnCdO us ing e-beam evaporation. The contact resistivity and thermal stability of Ti/A u (200Ã… /800Ã…) and T i/Al/Pt/Au (200Ã… /800Ã… /400Ã… /800Ã…) films were studi ed as a function of anneal ing temperature, using the transmission-line method (TLM) to assess the quality of the contacts. The lowest specific contact resistivity obtaine d for Ti/Al/Pt/Au was 2.3 10-4 -cm2 after annealing to 500Â°C and that for Ti/Au was 1.6 10-4 -cm2 after annealing to 450Â°C. The sheet resistivity decreased with increasing annealing temperature due to the formation of oxygen vacancies, as the O atoms migrate from the ZnCdO layer and bond to the metal. Sheet resistance can be achieved on the order of 103 / after annealing at 600Â°C. The lowest transfer resistance for Ti/Au was 6.77 -mm after annealing at 450Â°C and that for Ti/Al/Pt/Au was 7.23 -mm after annealing at 500Â°C. Furthermore, Ti/Au shows superior thermal stability to Ti/Al/Pt/A u with temperatures as high as 600Â°C. The film roughness of Ti/Al/P t/Au increased after annealing to 350Â°C, due to the outdiffusion of Al. This is undesirable b ecause the surface roughening may affect the edge dimensions of small device features. The dominant mechanism for current transport is field emission because the oxygen in the ZnCd O reacts with Ti in the metal scheme to form a thin TiOx layer with increasing annealing temperature. As TiOx is formed, oxygen vacancies occur in ZnCdO, which incr eases the carrier con centration. This enhances the tunneling probability. The specific contact resistivity depe nds on the barrier height and doping concentration. Therefore, a nother practical way to improve the contact resistivity is to increase the concentration of dopant within the semiconductor. This allows the carriers to tunnel through the barrier. Using pulse laser deposition, highly conductive aluminum-
129 doped ZnO samples were generated with a doping density on the order of ~1019 cm-3. The as-deposited contact had a low specific contact resistivity of 2.4 10-7 -cm2. The lowest contact resistivity found was using Ti/Au on alumi num-doped ZnO, which had a value of 6.0 10-8 -cm2 after annealing to 300Â°C. Th e sheet resistance of aluminumdoped ZnO is from 90 to 100 / . As the anneal progresses to 450Â°C, oxygen is transferred from the aluminum-doped ZnO to th e Ti/Au. This transfer is accompanied by surface roughening. The carrier transport m echanism for these contacts was dominated by the tunneling of an electron or a hole from the metal to the semiconductor. Indium-tin-oxide (ITO) has a refractive i ndex between that of ZnO and air, which reduces light reflection at the ZnO/air interface. ITO/Ti/A u (500Ã… /200Ã… /800Ã…) contacts were deposited on highly conduciv e aluminum-doped ZnO, and the circular transmission line method (c-TLM) was used to assess the quality of these contacts. The sheet resistance was on the order of 100 / . The contact resistivity of this metal stack ranges from 10-5 to 10-6 -cm2, with the morphology remaining thermally stable up to 350Â°C. Above 350Â°C, the oxygen is removed from the aluminum-doped ZnO, which leads to the decomposition of the aluminum -doped ZnO. This metal contact scheme appears to be a promising transparent conduc ting current spreading layer for ZnO-based LEDs. To understand carrier transport in hetero structures, it is important to have a detailed understanding of the conduction and va lence band offset. ZnCdO is an excellent material for the narrow bandgap active region. These results provi ded an energy diagram that will contribute to developing a ZnObased LED. The valence band offset of
130 Zn0.95Cd0.05O/ZnO (0001) was 0.17 eV, whereas the conduction band offset of this system was ~0.30 eV for a ZnCdO/ZnO heterostructure. The band offsets of MgO/GaN and Sc2O3/GaN heterostructures were also studied, as MgO and Sc2O3 are promising gate dielectrics and surface passivation films for GaN transistors. MgO/GaN has a valence band offset of 1.06 eV and a conduction band offset of 3.30 eV. These offsets show that th e MgO/GaN system is well suited for high temperature applications. The valence band offset of the Sc2O3/GaN heterostructure has a value of 0.42eV, and a conduction band offset of 2.56eV was obtained. These are also excellent values for high-temperature devi ce applications, which need strong carrier confinement. Using SiLENSe so ftware, p-ZnMgO/ZnCdO/n-ZnMgO (400nm/100nm/400nm) LED structures were simu lated. Device parameters, such as the doping concentration and semiconductor thickne ss, as well as the material properties were calculated in the simulation. The thickness and doping concentration of ZnCdO were optimized. The findings show that th e light intensity increases with increasing thickness and doping concentrati on of the ZnCdO active layer. In the parameter space explored, the optimum active laye r thickness was found to be 200 nm. To demonstrate a LED made from ZnO-ba sed materials, both pand n-type ZnO needs to be realized. Ohmic contacts must be developed on p-type ZnO-based material. This will include possible contact alloy formation using various metals, at optimal annealing temperatures. Another way to ma ke an ohmic contact is to make a tunnel contact by increasing the doping concentration to 1019cm-3. ZnMgO/ZnO and ZnMgO/ZnCdO heterostruct ure LEDs should be studied in the future. The band offset of these heterostruct ures can help us to design ZnO-based LEDs.
131 In addition, the measurements should focus on how the band offset changes on varying the concentrations of Mg and Cd within the lattice match condition for these heterostructures. This will allow the development of quantum well ZnO-based LED technology.
132 APPENDIX A PYSICAL PROPERTIES OF METALS Metal Ti Al Pt Au Electrical Resistivity ( -m) 3.90E-072.73E-081.08E-062.27E-08 Melting Point (oC) 1668 660.32 1768.4 1064.18 Work Function (eV) 4.33 4.2 5.64 5.47 Thermal Conductivity (W/cm K)0.219 2.37 0.716 3.17 Density (g/cm3) 4.51 2.7 21.5 19.3 Reference: .
133 APPENDIX B PHYSICAL PROPERTIES OF ZINC OXIDE Property ZnO Ref Crystal structure Wurtzite  Gap: direct (D)/indirect (I) D  Lattice constant (a0: Ã…) a0=3.250 c0=5.204  Bandgap energy (Eg: eV) 3.37  Intrinsic carrier concentration (ni: cm-3)<106  Electron mobility (Âµn: cm2/Vs) 200  Hole mobility (Âµp: cm2/Vs) ~10  Electron affinity ( : eV) 4.35 and 2.088[35,37] Electron effective mass (me*) 0.24me  Heavy hole effective mass (mhh*) 0.59me  Refractive index ( ) 2.008, 2.029  Exciton binding energy (meV) 60  Melting point (oC) 1975  Density (g/cm3) 5.7  Specific Heat (cal/goC) 0.125  Mohs Hardness 4  Thermal Expansion Coefficient (/K) 2.90x10-6  Thermal conductivity (cal/cm/k) 0.06  Dislocation density (cm-2) <100  Flat orientation <0001>  Linear expansion coefficient (/Â°C) a0: 6.5 x 10-6  Linear expansion coefficient (/Â°C) c0: 3.5 x 10-6  Static dielectric constant 8.656  Debye Temperature (Â°C) 143  Molecular Weight (g/mol) 81.39  Crystal-field splitting (meV) 42  Spin-orbit splitting (meV) -5  g-factors gn, c=gn, c -1.95  g-factors gp 0.8 
134 LIST OF REFERENCES 1. E. F. Schubert, Light-emitting diodes, Ca mbridge University Press, Cambridge and New York (2003). 2. M. Fukuda, Reliability and degradation of semiconductor lasers and LEDs, Artech House, Boston (1991). 3. E. F. Schubert, Light Emitting Diodes, http://www.ecse.rpi. edu/~schubert/LightEmitting-Diodes-dot-org/ (June 2006). 4. S. Nakamura, S. J. Pearton, and G. Fasol, The blue laser diode : the complete story, 2nd updated and extended ed, Springer, Berlin ; New York (2000). 5. S. Nakamura, T. Mukai, and M. Senoh, "High-brightne ss InGaN/AlGaN doubleheterostructure blue-green -light-emitting diodes," Jour nal of Applied Physics 76, 8189 (1994). 6. S. Nakamura, N. Iwasa, and M. Senoh, United States Patent No. 5,306,662, November 2 1992. 7. S. Nakamura, M. Senoh, N. Iwasa, a nd S. Nagahama, "High-power InGaN singlequantum-well-structure blue and violet light-emitting-diodes," Applied Physics Letters 67, 1868 (1995). 8. H. Sheng, N. W. Emanetoglu, S. Muthuku mar, B. V. Yakshinskiy, S. Feng, and Y. Lu, "Ta/Au ohmic contacts to n-type ZnO," Journal of Electronic Materials 32, 935 (2003). 9. H. K. Kim, S. H. Han, T. Y. Seong, and W. K. Choi, "Low-resistance Ti/Au ohmic contacts to Al-doped ZnO laye rs," Applied Physics Letters 77, 1647 (2000). 10. H. K. Kim, K. K. Kim, S. J. Park, T. Y. Seong, and I. Adesida, "Formation of low resistance nonalloyed Al/Pt ohmic contacts on n-type ZnO epitaxi al layer," Journal of Applied Physics 94, 4225 (2003). 11. Y. R. Ryu, S. Zhu, D. C. Look, J. M. Wrobel, H. M. Jeong, and H. W. White, "Synthesis of p-type ZnO films," Journal of Crystal Growth 216, 330 (2000). 12. S. S. Li, Semiconductor physical el ectronics, Plenum, New York (1993).
135 13. D. C. Look, D. C. Reynolds, C. W. L itton, R. L. Jones, D. B. Eason, and G. Cantwell, "Characterization of homoep itaxial p-type ZnO grown by molecular beam epitaxy," Applied Physics Letters 81, 1830 (2002). 14. D. C. Look, D. C. Reynolds, J. R. Sizel ove, R. L. Jones, C. W. Litton, G. Cantwell, and W. C. Harsch, "Electrical pr operties of bulk ZnO," Solid State Communications 105, 399 (1998). 15. A. Tsukazaki, A. Ohtomo, T. Onuma, M. Ohtani, T. Makino, M. Sumiya, K. Ohtani, S. F. Chichibu, S. Fuke, Y. Segawa, H. Ohno, H. Koinuma, and M. Kawasaki, "Repeated temperature modulat ion epitaxy for p-type doping and lightemitting diode based on ZnO," Nature Materials 4, 42 (2005). 16. A. Tsukazaki, M. Kubota, A. Ohtomo, T. Onuma, K. Ohtani, H. Ohno, S. F. Chichibu, and M. Kawasaki, "Blue light-emitting diode based on ZnO," Japanese Journal of Applied Physics Part 2-Letters & Express Letters 44, L643 (2005). 17. S. J. Jiao, Z. Z. Zhang, Y. M. Lu, D. Z. Shen, B. Yao, J. Y. Zhang, B. H. Li, D. X. Zhao, X. W. Fan, and Z. K. Tang, "ZnO p-n junction light-emitting diodes fabricated on sapphire substrat es," Applied Physics Letters 88 (2006). 18. K. Ip, Ph.D dissertation, University of Florida (2005). 19. S. J. Pearton, D. P. Norton, K. Ip, Y. W. Heo, and T. Steiner, "Recent progress in processing and properties of ZnO, " Progress in Materials Science 50, 293 (2005). 20. K. Ip, G. T. Thaler, H. S. Yang, S. Y. Han, Y. J. Li, D. P. Norton, S. J. Pearton, S. W. Jang, and F. Ren, "Contacts to ZnO," Journal of Crystal Growth 287, 149 (2006). 21. D. C. Look, "Recent advances in ZnO materials and devices, " Materials Science and Engineering B-Soli d State Materials for Advanced Technology 80, 383 (2001). 22. U. Ozgur, Y. I. Alivov, C. Liu, A. Teke , M. A. Reshchikov, S. Dogan, V. Avrutin, S. J. Cho, and H. Morkoc, "A compre hensive review of ZnO materials and devices," Journal of Applied Physics 98 (2005). 23. S. J. Pearton, D. P. Norton, K. Ip, Y. W. Heo, and T. Steiner, "Recent advances in processing of ZnO," Journal of Vacuum Science & Technology B 22, 932 (2004). 24. New Jersey Zinc Company and H. E. Br own, Zinc oxide rediscovered, New York (1957). 25. Y. W. Heo, K. Ip, S. J. Park, S. J. Pearton, and D. P. Norton, "Shallow donor formation in phosphorus-doped ZnO thin films," Applied Physics a-Materials Science & Processing 78, 53 (2004).
136 26. D. C. Look, C. Coskun, B. Claflin, a nd G. C. Farlow, "Electrical and optical properties of defects and impurities in ZnO," Physica B-Condensed Matter 340, 32 (2003). 27. J. D. Plummer, M.l Deal, and P. B. Griffin, Silicon VLSI technology : fundamentals, practice and modeling, Pr entice Hall, Upper Saddle River, NJ (2000). 28. S. A. Campbell, The science and engin eering of microelectronic fabrication, 2nd ed, Oxford University Press, New York (2001). 29. R. J. Shul, Plasma Etching and RIE: Fundamental and applications, AVS Short Course Program, Orlando, FL (2006). 30. K. Sangwal, Etching of crystals : th eory, experiment, and application, NorthHolland ;Sole distributors for the USA a nd Canada, Elsevier Science Pub. Co., Amsterdam ; New York (1987). 31. L. J. Brillson, Contacts to semiconduc tors : fundamentals and technology, Noyes, Park Ridge, N.J. (1993). 32. James W. Mayer and S. S. Lau, Electronic materials science : for integrated circuits in Si and GaAs, Macmillan ;Collier Macmillan, New York and London (1990). 33. S. M. Sze, Physics of semiconductor devices, 2nd ed, Wiley, New York (1981). 34. S. Mahajan and K. S. SreeHarsha, Pr inciples of growth and processing of semiconductors, WCB/McGr aw-Hill, Boston (1999). 35. Y. I. Alivov, E. V. Kalinina, A. E. Cherenkov, D. C. Look, B. M. Ataev, A. K. Omaev, M. V. Chukichev, and D. M. Bagna ll, "Fabrication and characterization of n-ZnO/p-AlGaN heterojunction light-em itting diodes on 6H-SiC substrates," Applied Physics Letters 83, 4719 (2003). 36. J. A. Aranovich, D. Golmayo, A. L. Fa hrenbruch, and R. H. Bube, "Photo-voltaic properties of ZnO-CdTe heterojunctions pr epared by spray pyrolysis," Journal of Applied Physics 51, 4260 (1980). 37. D. R. Lide, CRC handbook of chemistry and physics : a ready-reference book of chemical and physical data, 84th ed, CRC Press, Boca Raton (2003). 38. D. K. Schroder, Semiconductor materi al and device characterization, 2nd ed, Wiley, New York (1998). 39. G. K. Reeves and H. B. Harrison, "O btaining the specific co ntact resistance from transmission-line model measurements," Electron Device Letters 3, 111 (1982).
137 40. Department of Electrical & Computer Engineering Brigham Young University, Measuring Metal-Semiconductor Junc tion Contact Resistance Using the Transmission Line Method (TLM), http://www.ee.byu.edu/cleanroom/contact_resistance.phtml (June 2006). 41. G. K. Reeves, "Specific contact resi stance using a circular transmission-line model," Solid-State Electronics 23, 487 (1980). 42. J.H. Klootwijk and C.E. Timmering, presented at the Microelectronic Test Structures, 2004. Proceedings. ICMTS ' 04. The International Conference on, 2004 (unpublished). 43. E. A. Kraut, R. W. Grant, J. R. Waldrop, and S. P. Kowalczyk, "Precise determination of the valence-band ed ge in X-ray photoemission spectra application to measurement of semic onductor interface potentials," Physical Review Letters 44, 1620 (1980). 44. E. A. Kraut, R. W. Grant, J. R. Wa ldrop, and S. P. Kowalczyk, "Semiconductor core-level to valence-band maximu m binding-energy differences precise determination by X-ray photoelectronspectroscopy," Physical Review B 28, 1965 (1983). 45. R. W. Grant, E. A. Kraut, S. P. Kowa lczyk, and J. R. Waldrop, "Measurement of potential at semiconductor interfaces by el ectron-spectroscopy," Journal of Vacuum Science & Technology B 1, 320 (1983). 46. Y. Foulon, C. Priester, G. Allan, J. C. Garcia, and J. P. Landesman, "How does the chemical nature of the interface modify the band offset," Journal of Vacuum Science & Technology B 10, 1754 (1992). 47. S. K. Hong, T. Hanada, H. Makino, Y. F. Chen, H. J. Ko, T. Yao, A. Tanaka, H. Sasaki, and S. Sato, "Band alignment at a ZnO/GaN (0001) heterointerface," Applied Physics Letters 78, 3349 (2001). 48. T. Kitatani, M. Kondow, T. Kikawa, Y. Yazawa, M. Okai, and K. Uomi, "Analysis of band offset in GaNAs/GaAs by X-ray photoelectron spectroscopy," Japanese Journal of Applied Physics Part 1-Regular Papers Short Notes & Review Papers 38, 5003 (1999). 49. W. X. Ni and G. V. Hansson, "Band offsets in pseudomorphically grown Si/Si1xGex heterostructures studied with core -level X-ray photoele ctron-spectroscopy," Physical Review B 42, 3030 (1990). 50. W. X. Ni, J. Knall, and G. V. Hanss on, "New method to study band offsets applied to strained Si/Si1-xGex(100) heteroj unction interfaces," Physical Review B 36, 7744 (1987).
138 51. J. R. Waldrop and R. W. Grant, "M easurement of AlN/GaN (0001) heterojunction band offsets by x-ray photoemission sp ectroscopy," Applied Physics Letters 68, 2879 (1996). 52. T. E. Cook, C. C. Fulton, W. J. Mecouch, K. M. Tracy, R. F. Davis, E. H. Hurt, G. Lucovsky, and R. J. Nemanich, "Measurement of the band offsets of SiO2 on clean nand p-type GaN(0001)," Journal of Applied Physics 93, 3995 (2003). 53. G. Martin, A. Botchkarev, A. Ro ckett, and H. Morkoc, "Valence-band discontinuities of wurtzite GaN, AlN, a nd InN heterojunctions measured by x-ray photoemission spectroscopy," Applied Physics Letters 68, 2541 (1996). 54. C. C. Fulton, T. E. Cook, G. Lucovsky, and R. J. Nemanich, "Interface instabilities and electronic properties of ZrO2 on silicon (100)," J ournal of Applied Physics 96, 2665 (2004). 55. S. A. Chambers, T. Droubay, T. C. Kaspar, and M. Gutowski, "Experimental determination of valence band maxima for SrTiO3, TiO2, and SrO and the associated valence band offsets with Si (001)," Journal of Vacuum Science & Technology B 22, 2205 (2004). 56. L. Weinhardt, C. Heske, E. Umbach, T. P. Niesen, S. Visbeck, and F. Karg, "Band alignment at the i-ZnO/CdS interface in Cu (In,Ga)(S,Se)(2) thin -film solar cells," Applied Physics Letters 84, 3175 (2004). 57. R. Puthenkovilakam and J. P. Chang, "Valence band structure and band alignment at the ZrO2/Si interface," Applied Physics Letters 84, 1353 (2004). 58. L. F. Edge, D. G. Schlom, S. A. Ch ambers, E. Cicerrella, J. L. Freeouf, B. Hollander, and J. Schubert, "Measurement of the band offsets between amorphous LaAlO3 and silicon," Ap plied Physics Letters 84, 726 (2004). 59. H. Ohta, K. Kawamura, M. Orita, M. Hirano, N. Sarukura, and H. Hosono, "Current injection emission from a tr ansparent p-n junction composed of pSrCu2O2/n-ZnO," Applied Physics Letters 77, 475 (2000). 60. T. E. Cook, C. C. Fulton, W. J. Mecouc h, R. F. Davis, G. Lucovsky, and R. J. Nemanich, "Band offset measurements of the GaN (0001)/HfO2 interface," Journal of Applied Physics 94, 7155 (2003). 61. M. Oshima, S. Toyoda, T. Okumura, J. Okabayashi, H. Kumigashira, K. Ono, M. Niwa, K. Usuda, and N. Hirashita, "Chemist ry and band offsets of HfO2 thin films for gate insulators," Applied Physics Letters 83, 2172 (2003). 62. A. C. Tuan, T. C. Kaspar, T. Droubay, J. W. Rogers, and S. A. Chambers, "Band offsets for the epitaxial TiO2/SrTiO3/Si (001) system," Applied Physics Letters 83, 3734 (2003).
139 63. P. J. Hansen, V. Vaithyanathan, Y. Wu, T. Mates, S. Heikman, U. K. Mishra, R. A. York, D. G. Schlom, and J. S. Speck, "Rutile films grown by molecular beam epitaxy on GaN and AlGaN/GaN," Journa l of Vacuum Science & Technology B 23, 499 (2005). 64. G. V. Rao, F. Sauberlich, and A. Kl ein, "Influence of Mg content on the band alignment at CdS/(Zn,Mg)O interfaces," Applied Physics Letters 87 (2005). 65. T. Makino, Y. Segawa, M. Kawasaki, A. Ohtomo, R. Shiroki, K. Tamura, T. Yasuda, and H. Koinuma, "Band gap engineering based on MgxZn1-xO and CdyZn1-yO ternary alloy film s," Applied Physics Letters 78, 1237 (2001). 66. A. Franciosi and C. G. VandeWalle, "Heterojunction band offset engineering," Surface Science Reports 25, 1 (1996). 67. J. B. Hudson, Surface science : an introduction, John Wiley, New York (1998). 68. J. C. RiviÃ¨re and S. Myhra, Handbook of surface and interface analysis : methods for problem-solving, Marcel Dekker, New York (1998). 69. D. Briggs and M. P. Seah, Practical surface analysis, 2nd ed, Wiley ;Salle + SauerlÃ¤nder, Chichester ; New York and Aarau (1990). 70. C. R. Brundle, Charles A. Evans, and Shaun Wilson, Encyclopedia of materials characterization : surfaces, interfaces , thin films, Butterworth-Heinemann ;Manning, Boston and Greenwich, CT (1992). 71. R. Wiesendanger, Scanning probe mi croscopy and spectroscopy : methods and applications, Cambridge University Press, Cambridge ; New York (1994). 72. J. Goldstein, Scanning electron micr oscopy and x-ray microanalysis, 3rd ed, Kluwer Academic/Plenum Publishers, New York (2003). 73. J. F. Moulder, J. Chastain, and R. C. King, Handbook of x-ray photoelectron spectroscopy : a reference book of sta ndard spectra for identification and interpretation of XPS data, Physical El ectronics, Eden Prai rie, Minn. (1995). 74. H. Hartnagel, Semiconducting transparent thin films, Institute of Physics Pub., Bristol [England] ; Ph iladelphia, PA (1995). 75. P. Zu, Z. K. Tang, G. K. L. Wong, M. Kawasaki, A. Ohtomo, H. Koinuma, and Y. Segawa, "Ultraviolet spontaneous and stimulated emissions from ZnO microcrystallite thin films at room te mperature," Solid St ate Communications 103, 459 (1997). 76. D. M. Bagnall, Y. F. Chen, Z. Zhu, T. Yao, S. Koyama, M. Y. Shen, and T. Goto, "Optically pumped lasing of ZnO at room temperature," Applied Physics Letters 70, 2230 (1997).
140 77. M. Wraback, H. Shen, S. Liang, C. R. Goria, and Y. Lu, "High contrast, ultrafast optically addressed ultrav iolet light modulator base d upon optical anisotropy in ZnO films grown on R-plane sapph ire," Applied Physics Letters 74, 507 (1999). 78. J. M. Lee, K. K. Kim, S. J. Park, and W. K. Choi, "Low-re sistance and nonalloyed ohmic contacts to plasma treate d ZnO," Applied Physics Letters 78, 3842 (2001). 79. Y. F. Chen, D. Bagnall, and T. F. Yao, "ZnO as a novel photonic material for the UV region," Materials Science and Engi neering B-Solid State Materials for Advanced Technology 75, 190 (2000). 80. D. C. Look, J. W. Hemsky, and J. R. Sizelove, "Residual na tive shallow donor in ZnO," Physical Review Letters 82, 2552 (1999). 81. S. O. Kucheyev, J. E. Bradby, J. S. Williams, C. Jagadish, and M. V. Swain, "Mechanical deformation of single-cr ystal ZnO," Applied Physics Letters 80, 956 (2002). 82. D. C. Reynolds, D. C. Look, and B. Jogai, "Optically pumped ultraviolet lasing from ZnO," Solid State Communications 99, 873 (1996). 83. T. Aoki, Y. Hatanaka, and D. C. Look, "ZnO diode fabricated by excimer-laser doping," Applied Physics Letters 76, 3257 (2000). 84. C. C. Chang and Y. E. Chen, "Fabrica tion of high sensitivity ZnO thin film ultrasonic devices by electrochemical et ch techniques," Ieee Transactions on Ultrasonics Ferroelectrics and Frequency Control 44, 624 (1997). 85. P. M. Verghese and D. R. Clarke, "Pi ezoelectric contributions to the electrical behavior of ZnO varistors," Journal of Applied Physics 87, 4430 (2000). 86. C. R. Gorla, N. W. Emanetoglu, S. Lia ng, W. E. Mayo, Y. Lu, M. Wraback, and H. Shen, "Structural, optical, and surface acoustic wave pr operties of epitaxial ZnO films grown on (0112) sapphire by metalo rganic chemical vapor deposition," Journal of Applied Physics 85, 2595 (1999). 87. M. Joseph, H. Tabata, and T. Kawai, "P-type electrical c onduction in ZnO thin films by Ga and N codoping," Japanese Jour nal of Applied Physics Part 2-Letters 38, L1205 (1999). 88. S. Krishnamoorthy, A. A. Iliadis, A. Inumpudi, S. Choopun, R. D. Vispute, and T. Venkatesan, "Observation of resonant tunneling action in ZnO/Zn0.8Mg0.2O devices," Solid-State Electronics 46, 1633 (2002). 89. Y. Li, G. S. Tompa, S. Liang, C. Gorl a, Y. Lu, and J. Doyle, "Transparent and conductive Ga-doped ZnO films grown by lo w pressure metal organic chemical vapor deposition," Journal of Vacuum Sc ience & Technology a-Vacuum Surfaces and Films 15, 1063 (1997).
141 90. D. C. Look, D. C. Reynolds, J. W. He msky, R. L. Jones, and J. R. Sizelove, "Production and annealing of electron irradiation damage in ZnO," Applied Physics Letters 75, 811 (1999). 91. Y. R. Ryu, T. S. Lee, J. H. Leem, a nd H. W. White, "Fabrica tion of homostructural ZnO p-n junctions and ohmi c contacts to arsenic-doped p-type ZnO," Applied Physics Letters 83, 4032 (2003). 92. X. L. Guo, J. H. Choi, H. Tabata, an d T. Kawai, "Fabrication and optoelectronic properties of a transparent ZnO homostru ctural light-emitting diode," Japanese Journal of Applied Physics Part 2-Letters 40, L177 (2001). 93. Y. I. Alivov, J. E. Van Nostrand, D. C. Look, M. V. Chukichev, and B. M. Ataev, "Observation of 430 nm electrolumines cence from ZnO/GaN heterojunction lightemitting diodes," Applied Physics Letters 83, 2943 (2003). 94. H. Hosono, H. Ohta, K. Hayashi, M. Orita, and M. Hirano, "Near-UV emitting diodes based on a transparent p-n Junction composed of heteroepitaxially grown pSrCu2O2 and n-ZnO," Journal of Crystal Growth 237, 496 (2002). 95. A. Osinsky, J. W. Dong, M. Z. Kauser, B. Hertog, A. M. Dabiran, P. P. Chow, S. J. Pearton, O. Lopatiuk, and L. Chernyak, "MgZnO/AlGaN hete rostructure lightemitting diodes," Applied Physics Letters 85, 4272 (2004). 96. J. W. Dong, A. Osinsky, B. Hertog, A. M. Dabiran, P. P. Chow, Y. W. Heo, D. P. Norton, and S. J. Pearton, "Developmen t of MgZnO-ZnO-AlGaN heterostructures for ultraviolet light emitting applicati ons," Journal of Electronic Materials 34, 416 (2005). 97. D. K. Hwang, S. H. Kang, J. H. Lim, E. J. Yang, J. Y. Oh, J. H. Yang, and S. J. Park, "p-ZnO/n-GaN heterostructure ZnO light-emitting diodes," Applied Physics Letters 86 (2005). 98. A. Ohtomo, M. Kawasaki, I. Ohkubo, H. Koinuma, T. Yasuda, and Y. Segawa, "Structure and optical pr operties of ZnO/Mg0.2Zn0.8O superlattices," Applied Physics Letters 75, 980 (1999). 99. F. P. Koffyberg, "Thermoreflectance spectra of CdO band-gaps and bandpopulation effects," Physical Review B 13, 4470 (1976). 100. J. G. E. Gardeniers, Z. M. Rittersma, and G. J. Burger, "Preferred orientation and piezoelectricity in sputtered ZnO fi lms," Journal of Applied Physics 83, 7844 (1998). 101. H. Maki, T. Ikoma, I. Sakaguchi, N. Ohashi, H. Haneda, J. Tanaka, and N. Ichinose, "Control of surface morphol ogy of ZnO (0001) by hydrochloric acid etching," Thin Solid Films 411, 91 (2002).
142 102. Y. W. Heo, Y. W. Kwon, Y. Li, S. J. Pearton, and D. P. Norton, "P-type behavior in phosphorus-doped (Zn,Mg)O device stru ctures," Applied Physics Letters 84, 3474 (2004). 103. K. Ip, K. H. Baik, M. E. Overberg, E. S. Lambers, Y. W. Heo, D. P. Norton, S. J. Pearton, F. Ren, and J. M. Zavada, "Eff ect of high-density plasma etching on the optical properties and surface stoichiome try of ZnO," Applied Physics Letters 81, 3546 (2002). 104. I. S. Didenko, F. S. Hickernell, and N. F. Naumenko, "The experimental and theoretical characterization of the SAW propagation propert ies for zinc oxide films on silicon carbide," Ieee Tr ansactions on Ultrasonics Fe rroelectrics and Frequency Control 47, 179 (2000). 105. M. A. Contreras, K. Ramanathan, J. AbuShama, F. Hasoon, D. L. Young, B. Egaas, and R. Noufi, "Diode (characteristics in state-of-the-art ZnO/CdS/Cu(In(1x)Gax)Se-2 solar cells," Progress in Photovoltaics 13, 209 (2005). 106. A. Bergh, G. Craford, A. Duggal, a nd R. Haitz, "The promise and challenge of solid-state lighting," Physics Today 54, 42 (2001). 107. A. J. Nozik, "Photoelectrochemistry applications to sola r-energy conversion," Annual Review of Physical Chemistry 29, 189 (1978). 108. M. Gratzel, "Photoelect rochemical cells," Nature 414, 338 (2001). 109. D. C. Look, B. Claflin, Y. I. Alivov, and S. J. Park, "The future of ZnO light emitters," Physica Status Solidi a-Applied Research 201, 2203 (2004). 110. T. Makino, Y. Segawa, M. Kawasaki, and H. Koinuma, "Optical properties of excitons in ZnO-based quantum well hete rostructures," Semiconductor Science and Technology 20, S78 (2005). 111. S. H. Kim, H. K. Kim, and T. Y. Se ong, "Effect of hydrogen peroxide treatment on the characteristics of Pt Schottky contact on n-type ZnO," Applied Physics Letters 86 (2005). 112. T. Akane, K. Sugioka, and K. Midorikaw a, "Nonalloy ohmic contact fabrication in a hydrothermally grown n-ZnO (0001) substr ate by KrF excimer laser irradiation," Journal of Vacuum Science & Technology B 18, 1406 (2000). 113. H. K. Kim, J. W. Bae, K. K. Kim, S. J. Park, T. Y. Seong, and I. Adesida, "Inductively-coupled-plasma reactive i on etching of ZnO using BCl3-based plasmas and effect of the plasma treatme nt on Ti/Au ohmic contacts to ZnO," Thin Solid Films 447, 90 (2004). 114. A. A. Iliadis, R. D. Vispute, T. Venkate san, and K. A. Jones, "Ohmic metallization technology for wide band-gap semiconductors," Thin Solid Films 420, 478 (2002).
143 115. A. Inumpudi, A. A. Iliadis, S. Krishnam oorthy, S. Choopun, R. D. Vispute, and T. Venkatesan, "Pt-Ga ohmic contacts to n-Zn O using focused ion beams," Solid-State Electronics 46, 1665 (2002). 116. H. K. Kim, S. H. Han, T. Y. Seong, and W. K. Choi, "Elect rical and structural properties of Ti/Au ohmic contacts to n-ZnO," Journal of the Electrochemical Society 148, G114 (2001). 117. S. H. Kim, K. K. Kim, S. J. Park, a nd T. Y. Seong, "Thermally stable and low resistance Re/Ti/Au ohmic contacts to nZnO," Journal of the Electrochemical Society 152, G169 (2005). 118. H. K. Kim, I. Adesida, K. K. Kim, S. J. Park, and T. Y. Seong, "Study of the electrical and structural characteristics of Al/Pt oh mic contacts on n-type ZnO epitaxial layer," Journal of the Electrochemical Society 151, G223 (2004). 119. H. K. Kim, T. Y. Seong, K. K. Kim, S. J. Park, Y. S. Yoon, and I. Adesida, "Mechanism of nonalloyed Al ohmic contacts to n-type ZnO : Al epitaxial layer," Japanese Journal of Applied Physics Part 1-Regular Papers Sh ort Notes & Review Papers 43, 976 (2004). 120. C. T. Lee, H. W. Chen, and H. Y. Le e, "Metal-oxide-semiconductor devices using Ga2O3 dielectrics on n-type Ga N," Applied Physics Letters 82, 4304 (2003). 121. S. H. Kim, J. T. Maeng, C. J. Choi, J. H. Leem, M. S. Han, and T. Y. Seong, "Pt/indium tin oxide ohmic contacts to arsenic-doped p-type ZnO layers," Electrochemical and Solid State Letters 8, G167 (2005). 122. S. Kim, B. S. Kang, F. Ren, Y. W. Heo, K. Ip, D. P. Norton, and S. J. Pearton, "Contacts to p-type ZnMgO, " Applied Physics Letters 84, 1904 (2004). 123. H. S. Yang, S. Y. Han, K. H. Baik, S. J. Pearton, and F. Ren, "Effect of inductively coupled plasma damage on performance of GaN-InGaN multiquantum-well lightemitting diodes," Applied Physics Letters 86 (2005). 124. H. Sheng, N. W. Emanetoglu, S. Mut hukumar, S. Feng, and Y. Lu, "Nonalloyed Al ohmic contacts to MgxZn1-xO," Journal of Electronic Materials 31, 811 (2002). 125. S. Y. Kim, H. W. Jang, J. K. Kim, C. M. Jeon, W. I. Park, G. C. Yi, and J. L. Lee, "Low-resistance Ti/Al ohmi c contact on undoped ZnO," Journal of Electronic Materials 31, 868 (2002). 126. K. Ip, Y. W. Heo, K. H. Baik, D. P. No rton, S. J. Pearton, and F. Ren, "Carrier concentration dependence of Ti/Al/Pt/Au contact resistance on n-type ZnO," Applied Physics Letters 84, 544 (2004).
144 127. S. H. Kim, S. W. Jeong, D. K. Hwang, S. J. Park, and T. Y. Seong, "Zn/Au ohmic contacts on n-type ZnO epitaxial layers for light-emitting devices," Electrochemical and Solid State Letters 8, G198 (2005). 128. H. K. Kim, K. K. Kim, S. J. Park, T. Y. Seong, and Y. S. Yoon, "Thermally stable and low resistance Ru ohmic contacts to n-ZnO," Japanese Journal of Applied Physics Part 2-Letters 41, L546 (2002). 129. S. P. Jung, D. Ullery, C. H. Lin, H. P. L ee, J. H. Lim, D. K. Hwang, J. Y. Kim, E. J. Yang, and S. J. Park, "High-performance GaN-based light-emitting diode using high-transparency Ni/Au/Al-doped ZnO composite contacts," Applied Physics Letters 87 (2005). 130. D. W. Kim, H. Y. Lee, G. Y. Yeom, and Y. J. Sung, "A study of transparent contact to vertical GaN-based light-emitting diodes," Journal of Applied Physics 98 (2005). 131. J. K. Kim, T. Gessmann, E. F. Schubert, J. Q. Xi, H. Luo, J. Cho, C. Sone, and Y. Park, "GaInN light-emitting diode with c onductive omnidirectional reflector having a low-refractive-index indium-tin oxi de layer," Applied Physics Letters 88 (2006). 132. S. Y. Kim, H. W. Jang, and J. L. Lee, "High-brightness GaN-based light-emitting diode with indium tin oxide based transp arent ohmic contact," Journal of Vacuum Science & Technology B 22, 1851 (2004). 133. J. O. Song, W. K. Hong, Y. Park, J. S. Kwak, and T. Y. Seong, "Low-resistance Albased reflectors for high-power GaN-ba sed flip-chip light-emitting diodes," Applied Physics Letters 86 (2005). 134. W. K. Hong, J. O. Song, H. G. Hong, K. Y. Ban, T. Lee, J. S. Kwak, Y. Park, and T. Y. Seong, "Highly reflective and low resistance indium tin oxide/Ag ohmic contacts to p-type GaN for flip-chip light emitting diodes," Electrochemical and Solid State Letters 8, G320 (2005). 135. David R. Lide, CRC handbook of chemis try and physics : a ready-reference book of chemical and physical data, 76th ed, CRC Press, Boca Raton, FL (1995). 136. A. Y. C. Yu, "Electron tunneling and c ontact resistance of metal-silicon contact barriers," Solid-State Electronics 13, 239 (1970). 137. Y. M. Lu, W. S. Hwang, and J. S. Ya ng, "Effects of substrate temperature on the resistivity of non-stoichiometric sput tered NiOx films," Surface & Coatings Technology 155, 231 (2002). 138. L. M. Kukreja, S. Barik, and P. Mi sra, "Variable band gap ZnO nanostructures grown by pulsed laser deposition," Journal of Crystal Growth 268, 531 (2004).
145 139. K. Sakurai, T. Takagi, T. Kubo, D. Kajita , T. Tanabe, H. Takasu, S. Fujita, and S. Fujita, "Spatial composition fluctu ations in blue-luminescent ZnCdO semiconductor films grown by molecular beam epitaxy," Journal of Crystal Growth 237, 514 (2002). 140. E. Kohn, I. Daumiller, P. Schmid, N. X. Nguyen, and C. N. Nguyen, "Large signal frequency dispersion of Al GaN GaN heterostructure field effect transistors," Electronics Letters 35, 1022 (1999). 141. B. M. Green, K. K. Chu, E. M. Chumbes, J. A. Smart, J. R. Shealy, and L. F. Eastman, "The effect of surface passivat ion on the microwave characteristics of undoped AlGaN/GaN HEMT's," I eee Electron Device Letters 21, 268 (2000). 142. G. Simin, X. Hu, N. Ilinskaya, A. Kumar, A. Koudymov, J. Zhang, M. A. Khan, R. Gaska, and M. S. Shur, "7.5kW/mm(2) current switch using AlGaN/GaN metaloxide-semiconductor heterostructure field e ffect transistors on SiC substrates," Electronics Letters 36, 2043 (2000). 143. I. Daumiller, D. Theron, C. Gaquiere, A. Vescan, R. Dietrich, A. Wieszt, H. Leier, R. Vetury, U. K. Mishra, I. P. Smorchkova, S. Keller, N. X. Nguyen, C. Nguyen, and E. Kohn, "Current inst abilities in GaN-based de vices," Ieee Electron Device Letters 22, 62 (2001). 144. P. B. Klein, S. C. Binari , K. Ikossi, A. E. Wickenden, D. D. Koleske, and R. L. Henry, "Current collapse and the role of carbon in AlGaN/GaN high electron mobility transistors grown by metalorganic vapor-phase epitaxy," Applied Physics Letters 79, 3527 (2001). 145. S. C. Binari, K. Ikossi, J. A. Roussos, W. Kruppa, D. Park, H. B. Dietrich, D. D. Koleske, A. E. Wickenden, and R. L. Henry, "Trapping effects and microwave power performance in AlGaN/GaN HE MTs," Ieee Transactions on Electron Devices 48, 465 (2001). 146. L. F. Eastman, V. Tilak, J. Smart, B. M. Green, E. M. Chumbes, R. Dimitrov, H. Kim, O. S. Ambacher, N. Weimann, T. Prunt y, M. Murphy, W. J. Schaff, and J. R. Shealy, "Undoped AlGaN/GaN HEMTs for microwave power amplification," Ieee Transactions on Electron Devices 48, 479 (2001). 147. B. Luo, J. W. Johnson, J. Kim, R. M. Me handru, 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, "Influence of MgO and Sc2O 3 passivation on AlGa N/GaN high-electronmobility transistors," Applied Physics Letters 80, 1661 (2002). 148. A. V. Vertiatchikh, L. F. Eastman, W. J. Schaff, and T. Prunty, "Effect of surface passivation of AlGaN/GaN hete rostructure field-effect transistor," Electronics Letters 38, 388 (2002).
146 149. M. A. Khan, X. Hu, G. Sumin, A. Lune v, J. Yang, R. Gaska, and M. S. Shur, "AlGaN/GaN metal oxi de semiconductor heterostructur e field effect transistor," Ieee Electron Device Letters 21, 63 (2000). 150. M. A. Khan, X. Hu, A. Tarakji, G. Simin, J. Yang, R. Gaska, and M. S. Shur, "AlGaN/GaN metal-oxide-sem iconductor heterostructure fi eld-effect transistors on SiC substrates," Applied Physics Letters 77, 1339 (2000). 151. 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, "Comparison of surface passivation films for reduction of current collapse in AlGaN/GaN high electron mobility transistors," Journal of the Electrochemical Society 149, G613 (2002). 152. B. Luo, R. M. Mehandru, J. Kim, F. Ren, B. P. Gila, A. H. Onstine, C. R. Abernathy, S. J. Pearton, R. C. Fitch, J. Gillespie, R. Dellmer, T. Jenkins, J. Sewell, D. Via, and A. Crespo, "The role of cleaning conditions and epitaxial layer structure on reliability of Sc2O3 and MgO passivation on AlGaN/GaN HEMTS," Solid-State Electronics 46, 2185 (2002). 153. S. C. Binari, K. Doverspike, G. Kelner, H. B. Dietrich, and A. E. Wickenden, "GaN FETs for microwave and high-temperature applications," Solid-State Electronics 41, 177 (1997). 154. J. K. Gillespie, R. C. Fitch, J. Sewell, R. Dettmer, G. D. Via, A. Crespo, T. J. Jenkins, B. Luo, R. Mehandru, J. Kim, F. Ren, B. P. Gila, A. H. Onstine, C. R. Abernathy, and S. J. Pearton, "Effects of Sc2O3 and MgO passivation layers on the output power of AlGaN/GaN HEMTs, " Ieee Electron Device Letters 23, 505 (2002). 155. R. Mehandru, B. Luo, J. Kim, F. Ren, B. P. Gila, A. H. Onstine, C. R. Abernathy, S. J. Pearton, D. Gotthold, R. Birkhahn, B. Peres, R. Fitch, J. Gillespie, T. Jenkins, J. Sewell, D. Via, and A. Crespo, "AlGaN/GaN metal-oxi de-semiconductor high electron mobility transistors using Sc2O3 as the gate oxide and surface passivation," Applie d Physics Letters 82, 2530 (2003). 156. J. Kim, R. Mehandru, B. L uo, F. Ren, B. P. Gila, A. H. Onstine, C. R. Abernathy, S. J. Pearton, and Y. Irokawa, "Charact eristics of MgO/GaN gate-controlled metaloxide-semiconductor diodes," Applied Physics Letters 80, 4555 (2002). 157. B. P. Gila, A. H. Onstine, J. Kim, K. K. Allums, F. Ren, C. R. Abernathy, and S. J. Pearton, "Magnesium oxide gate diel ectrics grown on GaN using an electron cyclotron resonance plasma," Journa l of Vacuum Science & Technology B 21, 2368 (2003). 158. B. P. Gila, F. Ren, and C. R. Abernat hy, "Novel insulators for gate dielectrics and surface passivation of GaN-based elect ronic devices," Materials Science & Engineering R-Reports 44, 151 (2004).
147 159. G. Simin, A. Koudymov, A. Tarakji, X. Hu, J. Yang, M. A. Khan, M. S. Shur, and R. Gaska, "Induced strain mechanism of current collapse in AlGaN/GaN heterostructure field-effect tran sistors," Applied Physics Letters 79, 2651 (2001). 160. Y. Nakano, T. Kachi, and T. Jimbo, "C haracteristics of SiO2/n -GaN interfaces with beta-Ga2O3 interlayers," Applied Physics Letters 83, 4336 (2003). 161. C. Bae, C. Krug, G. Lucovsky, A. Chakraborty, and U. Mishra, "Work-function difference between Al and n-GaN from Al -gated n-GaN/nitrid ed-thin-Ga2O3/SiO2 metal oxide semiconductor structur es," Applied Physics Letters 84, 5413 (2004). 162. K. Matocha and R. J. Gutmann, "H igh-voltage normally off GaN MOSFETs on sapphire substrates," Ieee Tr ansactions on Electron Devices 52, 6 (2005). 163. Y. Irokawa, Y. Nakano, M. Ishiko, T. Kach i, J. Kim, F. Ren, B. P. Gila, A. H. Onstine, C. R. Abernathy, S. J. Pearton, C. C. Pan, G. T. Chen, and J. I. Chyi, "MgO/p-GaN enhancement mode meta l-oxide semiconductor field-effect transistors," Applied Physics Letters 84, 2919 (2004). 164. M. Higashiwaki, T. Matsui, and T. Mi mura, "AlGaN/GaN MIS HFETs with f(T) of 163 GHz using Cat-CVD SiN ga te-insulating and passivati on layers," Ieee Electron Device Letters 27, 16 (2006). 165. J. Kim, R. Mehandru, B. L uo, F. Ren, B. P. Gila, A. H. Onstine, C. R. Abernathy, S. J. Pearton, and Y. Irokawa, "Inversi on behavior in Sc2O3/GaN gated diodes," Applied Physics Letters 81, 373 (2002). 166. A. Y. Polyakov, N. B. Smirnov, A. V. Govorkov, V. N. Danilin, T. A. Zhukova, B. Luo, F. Ren, B. P. Gila, A. H. Onstine, C. R. Abernathy, and S. J. Pearton, "Deep traps in unpassivated and Sc2O3-passi vated AlGaN/GaN high electron mobility transistors," Applied Physics Letters 83, 2608 (2003). 167. J. J. Chen, B. P. Gila, M. Hlad, A. Ge rger, F. Ren, C. R. Abernathy, and S. J. Pearton, "Determination of MgO/GaN heterojunction band offsets by x-ray photoelectron spectro scopy," Applied Physics Letters 88 (2006). 168. J. R. Waldrop, R. W. Grant, S. P. Ko walczyk, and E. A. Kraut, "Measurement of semiconductor heterojunction band di scontinuities by X-ray photoemission spectroscopy," Journal of Vacuum Scie nce & Technology a-Vacuum Surfaces and Films 3, 835 (1985). 169. S. A. Ding, S. R. Barman, K. Horn, H. Yang, B. Yang, O. Brandt, and K. Ploog, "Valence band discontinuity at a cubic GaN/GaAs heterojunction measured by synchrotron-radiation photoemission sp ectroscopy," Applied Physics Letters 70, 2407 (1997).
148 170. J. H. Edgar, S. Strite, I. Akasaki, H. Amano, and C. Wetzel, Properties, Processing and Applications of Galliu m Nitride and Related Semiconductors, London, U.K. : IEE, INSPEC, c1999., (1999). 171. G. Coli and K. K. Bajaj, "Excitoni c transitions in ZnO/MgZnO quantum well heterostructures," Applied Physics Letters 78, 2861 (2001). 172. B. P. Gila, M. Hlad, A. H. Onstine, R. Frazier, G. T. Thaler, A. Herrero, E. Lambers, C. R. Abernathy, S. J. Pearton, T. Anderson, S. Jang, F. Ren, N. Moser, R. C. Fitch, and M. Freund, "Improve d oxide passivation of AlGaN/GaN high electron mobility transistor s," Applied Physics Letters 87 (2005). 173. J. Piprek, Optoelectronic devices: adva nced simulation and analysis, Springer, New York (2005). 174. Inc. Semiconductor Technology Research, Simulator for Light Emitters based on Nitride Semiconductors, http://www.semitech.us/products/SiLENSe/ (June 2006). 175. National Compound Semiconductor Roadmap, Compound Semiconductor Materials, http://www.onr.navy.mil/sci_t ech/31/312/ncsr/materials/ (June 2006). 176. D.W. Palmer, The Semiconductors-Information, http://www.semiconductors.co.uk/ (June 2006). 177. H. Karzel, W. Potzel, M. Kofferlein, W. Schiessl, M. Steiner, U. Hiller, G. M. Kalvius, D. W. Mitchell, T. P. Das, P. Blaha, K. Schwarz, and M. P. Pasternak, "Lattice dynamics and hyperfin e interactions in ZnO and ZnSe at high external pressures," Physical Review B 53, 11425 (1996). 178. Y. B. Jin, B. Zhang, S. M. Yang, Y. Z. Wang, J. Chen, H. H. Zhang, C. H. Huang, C. Q. Cao, H. Cao, and R. P. H. Chang, "Room temperature UV emission of MgxZn1-xO films," Solid State Communications 119, 409 (2001). 179. Molecular Technology, Singl e Crystalline Substrates, http://www.mtberlin.com/frames_cryst/descriptions/substrates.htm (June 2006). 180. Inc. Automation Creations, Ma tWeb Material Property Data, http://www.matweb.com/ (June 2006). 181. K. H. Hellwege, R. BÃ¶rnstein, and H. Landolt, Zahlenwerte und Funktionen aus Naturwissenschaften und Technik, neue Serie, Springer-Verlag, Berlin (1961).
149 BIOGRAPHICAL SKETCH Jau-Jiun Chen was born on January 2, 1977, in Taipei, Taiwan. She is the daughter of Su-Chu and Chi-Hua Chen. After gradua ting from Affiliated Se nior High School of National Taiwan Normal University in June 1995, she enrolled at the Department of Chemical Engineering at National Taiw an University in 1996. She conducted undergraduate research on the topic of th e scaling up mesoporous molecular sieves MCM-41 under the supervision of Prof. BenZu Wan. In June 2000, she earned her bachelorÂ’s degree in Chemical Engineering. She continued working as a research assistant under Prof. Ben-Zu Wan in August 2000. In August 2001, Jau-Jiun began her gradua te studies in chemi cal engineering at University of Florida. For the first three ye ars, she worked in the surface science field assisting with establishing a surface science la boratory. She also worked with the senior graduate student on the oxidida tion of Pt(111) and the oxidati on of nitrided Si(100). She earned her masterÂ’s degree in the fall of 2004. She then redirected her research efforts in the semiconductor field with emphasis on LED device formation. She studied for her doctorate under her academic advisors, Prof. Fan Ren and Prof. Jason F. Weaver. While in Gainesville, she has volunteered her time to help homeless and developmentally disabled people with the Ga inesville Buddhist Asso ciation. She also served as Vice President in the Taiwanese Stude nt Association at UF. One of the primary thrusts of her work as Vice President was to assist with new international student accommodation to the US. In her spare time she loves to listen to music, play basketball,
150 and run sprints. Her talents are not only limited to the field of science, but to music as well. She is proficient with Liu-Yeh Ch in, which is a Chinese banjo, piano and accordion.