<%BANNER%>

Silicon Carbide Schottky and P-I-N Rectifiers


PAGE 1

SILICON CARBIDE SCHOTTKY AND P-I-N RECTIFIERS By SAURAV NIGAM A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE UNIVERSITY OF FLORIDA 2003

PAGE 2

Copyright 2003 by Saurav Nigam

PAGE 3

To my parents, my fiance, my brother and my friends for their endless love and support.

PAGE 4

ACKNOWLEDGMENTS The past year has been one of the most crucial years of my life because for the first time in my professional career, I was introduced to an exciting realm of microelectronics. I would like to thank my advisor, Prof. Fan Ren, for giving me this golden opportunity to join his research group. His commitment to provide world-class facilities in his laboratory, engage students in the cutting-edge research and above all to provide excellent guidance in all the spheres of life, both professional and personal, is truly exceptional. Needless to mention, his extreme hard work and determination will always act as a source of inspiration to me and instill in me the confidence to face every adversity in life. He has always been able to find some time out of his extremely busy schedule to help me to steer my efforts in the right direction. I will always look forward to him as my true mentor. I am also indebted to the other members of my advisory committee: Steve Pearton and Tim Anderson. Their significant contributions to this work are greatly appreciated. During the collaboration with Prof. Steve Peartons and Prof. Cammy Abernathys research groups, I had an opportunity to develop interpersonal skills, emphasizing team efforts and it was here that I came across some of the most remarkable people, like Brent Gila and Mark, that I have ever known. I would like to thank Brent for always lending me a helping hand whenever I needed him and for his perspicacious views especially during the Wednesday meetings. I iv

PAGE 5

will also always cherish nice little discussions with Mark, who has provided me his considerable time and energy to help me to set up the dicing saw. I would also like to thank Ben Luo not just for providing the guidance in research but also for providing his invaluable advice and guidance. I will always cherish those moments when he would come up with his nice little humorous comments that would take away all the stress after the days work. I would always count on him as a truly sincere friend to seek for the right guidance. I would also like to thank Rishabh for his support and care. He has been patient in providing me his valuable time to gain hands-on experience with various equipment together with some delicious food and recipes. I have also learned a lot from Jihyun, who has collaborated most closely with my research. I would also like to thank Yoshi for providing me help and support and always showing a great deal of concern. His friendship is one of the greatest gifts to me. Kelly, Kwang and Kyu-pil have always been kind enough to lend their helping hand with the research. I would like to thank Dennis Vince and Jim Hinnat of the Chemical Engineering Departments machine shop for all their help. I would also like to thank the staff in the Chemical Engineering Department for all their administrative help. They are Peggy-Jo Daugherty, Sonja Pealer, Santiago, Nancy Krell, Janice Harris and Debbie Sadoval. Their assistance allowed me to focus on research. I would like to thank my parents, my brother and my fiance for their endless love, care and support. They have provided me the strength and inspiration to continue my studies at the University of Florida. This acknowledgement cannot be complete without thanking my roommates Gopal, Nori, Sudeep and Archit, who have been very understanding and caring throughout my stay at Gainesville. v

PAGE 6

TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES...........................................................................................................viii LIST OF FIGURES...........................................................................................................ix ABSTRACT......................................................................................................................xii CHAPTER 1 INTRODUCTION.........................................................................................................1 1.1 Background.............................................................................................................1 1.2 Current Status of SiC Power Technology...............................................................4 1.3 SiC Semiconductor Devices for Power Applications.............................................7 1.3.1 Material Availability and Quality...................................................................8 1.3.2 SiC Processing Technology............................................................................9 1.3.2.1 Ion implantation...................................................................................10 1.3.2.2 Oxides on SiC......................................................................................11 1.3.2.3 Contacts................................................................................................12 1.3.2.4 Edge termination and passivation........................................................13 2 SILICON CARBIDE SCHOTTKY RECTIFIERS.....................................................14 2.1 Experimental Characteristics of 4H-SiC Schottky Rectifiers...............................14 2.1.1 Introduction...................................................................................................14 2.1.2 Experimental Methods..................................................................................15 2.1.3 Results and Discussion..................................................................................17 2.2 Ni Contacts to N-Type 4H-SiC.............................................................................26 2.3 Stability of SiC Schottky Rectifiers to Rapid Thermal Annealing.......................27 2.3.1. Introduction..................................................................................................27 2.3.2 Experimental Methods..................................................................................29 2.3.3 Results and Discussion..................................................................................29 2.3.4 Summary and Conclusions............................................................................32 3 IRRADIATION AND PASSIVATION EFFECTS ON 4H-SiC RECTIFERS...........38 3.1 High Energy Proton Irradiation Effects on SiC Schottky Rectifiers....................38 3.1.1 Introduction...................................................................................................38 vi

PAGE 7

3.1.2 Experimental Methods..................................................................................39 3.1.3 Results and Discussion..................................................................................39 3.2 Influence of PECVD of SiO2 Passivation Layers on 4H-SiC Schottky Rectifiers41 3.2.1 Introduction...................................................................................................41 3.2.2 Experimental Methods..................................................................................45 3.2.3 Results and Discussion..................................................................................45 3.3 Effect on 4H-SiC Schottky Rectifier of Ar Discharges Generated in a Planar Inductively Coupled Plasma Source.....................................................................47 3.3.1 Introduction...................................................................................................47 3.3.2 Experimental Methods..................................................................................51 3.3.3 Results and Discussion..................................................................................51 3.3.4 Summary and Conclusions............................................................................54 4 JUNCTION TERMINATION EXTENSION GEOMETRY OF SiC RECTIFIERS..61 4.1 Influence of Edge Termination Geometry on Performance of 4H-SiC P-i-N Rectifiers...............................................................................................................62 4.1.1 Introduction...................................................................................................62 4.1.2 Experimental Methods..................................................................................63 4.1.3 Results and Discussion..................................................................................63 4.2 Effect of Contact Geometry on 4H-SiC Schottky Rectifiers with Junction Termination Extension..........................................................................................65 4.2.1 Introduction...................................................................................................65 4.2.2 Experimental Methods..................................................................................65 4.2.3 Results and Discussion..................................................................................72 4.3 Role of Device Area, Mesa Length and Metal Overlap Distance on Breakdown Voltage of 4H-SiC P-i-N Rectifiers......................................................................74 4.3.1 Introduction...................................................................................................74 4.3.2 Experimental Methods..................................................................................74 4.3.3 Results and Discussion..................................................................................82 4.3.4 Summary and Conclusions............................................................................84 LIST OF REFERENCES...................................................................................................93 BIOGRAPHICAL SKETCH...........................................................................................102 vii

PAGE 8

LIST OF TABLES Table page 1-1 Physical properties of important semiconductors for high voltage power devices..5 1-2 Normalized unipolar figures of merit of important semiconductors for high voltage power devices..............................................................................................5 viii

PAGE 9

LIST OF FIGURES Figure page 1-1 Comparison of the ideal breakdown voltage of Si and SiC devices for different doping levels............................................................................................................2 1-2 Comparison of the blocking layer thickness as a function of the ideal breakdown voltage for SiC and Si..............................................................................................3 2-1 Cross-section of SiC Schottky rectifiers (top) and photograph of completed 2 inch diameter wafer (bottom).................................................................................16 2-2 Reverse I-V characteristics at 25 0C from Schottky rectifiers with different contact diameters...................................................................................................18 2-3 Reverse I-V characteristic from 154 m diameter rectifier (top) and map of reverse breakdown voltage from a quarter of a 2 inch diameter wafer (bottom)..19 2-4 Comparison of breakdown voltage obtained from simulations and the fabricated devices....................................................................................................................21 2-5 Calculated reverse breakdown voltage as a function of epilayer doping...............23 2-6 Forward I-V characteristics in linear (top) and log (bottom) scales.......................24 2-7 Specific on-resistance as a function of breakdown voltage....................................25 2-8 Contact resistivity of Ni ohmic contact on n+ 4H-SiC as a function of annealing temperature (top), annealing time (center) and annealing ambient (bottom)........28 2-9 Forward I-V characteristics as a function of anneal temperature for 60 sec anneals....................................................................................................................33 2-10 Change in RON as a function of anneal temperature for 60 sec anneals.................34 2-11 Forward I-V characteristics (top) and change in VF (center) and RON (bottom) as a function of anneal time at 900C....................................................................35 2-12 Optical micrographs of Ni contacts before (top) and after annealing at 1000C for 60 sec (center) or 120 sec (bottom)..................................................................36 ix

PAGE 10

2-13 Change in VF (top) and RON (bottom) as a function of measurement temperature after 900C, 60 sec anneals....................................................................................37 3-1 Reverse I-V characteristics from 4H-SiC Schottky rectifiers before and after proton irradiation at a dose of 5 x 109 cm-2 (top) and percentage increase in reverse leakage at V (bottom)........................................................................42 3-2 Forward I-V characteristics from 4H-SiC Schottky rectifiers before and after proton irradiation at a dose of 5 x 109 cm-2 (top) and percentage decrease in forward current at 2V (bottom)..............................................................................43 3-3 Numerical change in values of n (top), RON (center) and VF (bottom) as a function of proton fluence......................................................................................44 3-4 Percentage change in VB as a function of plasma power (top), process pressure (center) and N2O content (bottom) during SiO2 deposition...................................48 3-5 Percentage change in VF as a function of plasma power (top) and process pressure (bottom) during SiO2 deposition..............................................................49 3-6 Percentage change in RON as a function of plasma power (top), N2O content (center) and process pressure (bottom) during SiO2 deposition............................50 3-7 Planar coil ICP reactor............................................................................................56 3-8 Percentage change in VB (top) and RON (bottom) of 4H-SiC rectifiers as a function of ICP source power................................................................................57 3-9 Percentage change in VF (top) and n (bottom) of 4H-SiC rectifiers as a function of ICP source power...............................................................................................58 3-10 Percentage change in VB (top) and RON (bottom) of 4H-SiC rectifiers as a function of process pressure...................................................................................59 3-11 Percentage change in VF (top) and n (bottom) of 4H-SiC rectifiers as a function of process pressure.................................................................................................60 4-1 P-i-N rectifier..........................................................................................................66 4-2 Forward I-V characteristics of p-i-n rectifier.........................................................67 4-3 Reverse I-V characteristics from p-i-n rectifiers as a function of metal overlap distance onto mesa.................................................................................................68 4-4 Variation of VB with metal overlap distance for p-i-n rectifiers............................69 4-5 Reverse I-V characteristics from p-i-n rectifiers as a function of mesa length......70 x

PAGE 11

4-6 Variation of VB with mesa length for p-i-n rectifiers with fixed metal overlap distance of 4 m.....................................................................................................71 4-7 Schottky rectifier....................................................................................................75 4-8 Forward I-V characteristics of 0.64 mm2 rectifier..................................................76 4-9 Reverse I-V characteristics of 0.04 mm2 rectifiers with different contact shape...77 4-10 Reverse I-V characteristics of rectifiers with oval shaped contacts of different area.........................................................................................................................78 4-11 Reverse I-V characteristics of rectifiers with oval shaped contacts (0.04 mm2), as a function of extent of metal overlap.................................................................79 4-12 Reverse I-V characteristics of rectifiers with oval-shaped contacts (0.04 mm2), as a function of extent of metal overlap.................................................................80 4-13 Variation of VB with extent of metal overlap for different JTE lengths.................81 4-14 Reverse I-V characteristics of p-i-n rectifiers as a function of active area.............86 4-15 VB as a function of diode area for p-i-n rectifiers with mesa length 20 m and metal overlap 4 m................................................................................................87 4-16 Reverse I-V characteristics of p-i-n rectifiers as a function of metal overlap distance..................................................................................................................88 4-17 VB as a function of metal overlap length for p-i-n rectifiers with area 0.04 mm2 and zero mesa length..............................................................................................89 4-18 Reverse I-V characteristics of p-i-n rectifiers as a function of mesa length...........90 4-19 VB as a function of mesa length for p-i-n rectifiers with area 0.04 mm2 and zero metal overlap length...............................................................................................91 4-20 Forward I-V characteristics....................................................................................92 xi

PAGE 12

Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science SILICON CARBIDE SCHOTTKY AND P-I-N RECTIFIERS By Saurav Nigam May 2003 Chair: Fan Ren Major Department: Chemical Engineering There has been a significant research interest in silicon carbide (SiC) over the past few years as a base material system for high frequency and high power semiconductor devices. SiC Schottky diodes have shown an excellent performance for the realization of power diodes for fast switching applications with nearly negligible power loss. In spite of extremely high switching speeds, Schottky diodes suffer from high leakage current. P-i-N diodes offer low leakage currents but show reverse recovery charge during switching and have a large junction forward voltage drop due to the wide bandgap of 4H-SiC. However, both the diodes have shown a considerable improvement in maximum breakdown voltage values compared to that of planar junctions, with the employment of proper edge termination methods. The dc performance of SiO2 dielectric overlap-terminated 4H-SiC Schottky rectifiers was simulated for different drift layer doping levels and dielectric thickness. The devices were then fabricated with different Schottky contact diameters and the dc xii

PAGE 13

results compared to the simulations. There was a strong dependence of reverse breakdown voltage (VB) on contact diameter () ranging from V for 100 m to 440 V for 1000 m. The reverse leakage current scaled with diode diameter, indicating that the surface contributions are dominant. The specific on-state resistance was ~0.36 mcm2, close to the theoretical minimum, with a forward turn-on voltage of 1.82 V at 100 A.cm-2. The contact resistivity of back-side Ni contacts annealed at 970 0C was ~1.2 x 10-6 cm2. The effect of deposition conditions of plasma enhanced chemical vapor deposition (PECVD) SiO2 layers and inductively coupled Ar plasmas on the electrical properties of 4H-SiC Schottky rectifiers was also studied. For the former, the changes in reverse breakdown voltage (VB), forward turn-on voltage (VF) and on-state resistance (RON) were 20% over a broad range of plasma conditions and show that 4H-SiC is relatively resistant to changes induced by the ion bombardment and hydrogen flux present during PECVD of dielectrics to surface passivation. In the latter, the VB increased by increase in both incident ion energy and ion flux. 4H-SiC P-i-N rectifiers with SiO2 passivated mesa edge termination showed forward current characteristics dominated by recombination at low bias (n~1.97) and diffusion at high voltages (n~1.1). The forward turn-on voltage was ~4V, with a specific on-state resistance of 15 mcm2, on/off current ratio of 1.5x105 at 3V/-450V and figure-of-merit, VB2/RON, of 13.5 MWcm-2. The mesa extension distance did not have a strong impact on reverse breakdown voltage. xiii

PAGE 14

CHAPTER 1 INTRODUCTION 1.1 Background Silicon carbide (SiC) has shown outstanding potential for high power, high temperature, high frequency and high voltage devices used in advanced communications and radar system sensor and control systems, utility power switching and traction motor control. The material has excellent physical and electrical properties such as extremely high thermal conductivity (up to 4.9 W/cmK), high breakdown electric field (2-4 x 106 V/cm), a wide bandgap (3.26 eV for 4H-polytype) and high electron saturation velocity (2 x 107 cm/s) as shown in Table 1-1 [1]. These characteristics result in significantly higher current densities, operating temperatures and breakdown voltage than comparable Si devices and in lower switching and on-state losses. A comparison of ideal breakdown voltages versus blocking layer doping concentration (Figure 1-1) suggests that the more highly doped blocking layer (more than 10 times higher) provides lower resistance for SiC because more majority carriers are present than for comparably rated Si devices. A comparison of voltage blocking layer thickness for a given breakdown voltage (Figure 1-2) shows a thinner blocking layer (n-~5 x 1015 cm-3) of SiC devices (1/10th that of Si devices) also contribute to the lowering of specific on-resistance by a factor of 10. The combination of 1/10th the blocking layer thickness with 10 times the doping concentration can yield a SiC device with a factor of 100 advantage in resistance compared to that of Si devices. 1

PAGE 15

2 101410151016101102103104 SiCSi Ideal breakdown voltage (V)Doping of the blocking layer (cm-3) Figure 1-1 Comparison of the ideal breakdown voltage of Si and SiC devices for different doping levels.

PAGE 16

3 103104100101102103 SiCSi Width of voltage blocking layer (m)Ideal breakdown voltage (V) Figure 1-2 Comparison of the blocking layer thickness as a function of the ideal breakdown voltage for SiC and Si.

PAGE 17

4 Several figures of merit, specifically to quantify the intrinsic performance potentials of unipolar and bipolar devices, have been proposed [2-5]. The most important material parameter is the avalanche electric field, followed by thermal conductivity and carrier mobility. The unipolar figures of merit are shown in Table 1-2 [1] where SiC and AlN unipolar devices have been projected to have intrinsic device improvement of 80 to 30000 times over conventional Si [4]. Also, bipolar devices, such as pin junction rectifier and GTO thyristors, are expected to have lower total power loss when the switching frequencies exceed 1 and 30 kHz respectively [5]. Here, it is worth pointing out that BN is the best among the Group III-nitrides for power devices because of its high thermal conductivity and indirect bandgap. Furthermore, due to the higher carrier mobility and more isotropic nature of its properties, 4H-SiC has become the polytype of choice [6]. 1.2 Current Status of SiC Power Technology The overall goal for power electronic circuits is to reduce power losses, volume, weight, and, costs of the system [7]. These factors become significant for applications in future military and commercial ships especially electric propulsion. The first SiC based device was a light emitting diode fabricated in 1907 by H. J. Round. However, a more intensive silicon carbide semiconductor material and device development has been underway for nearly 35 years in United States, the former Soviet Union, Europe and Asia. Only in the past 15 years with the input of substantial funding from the US DOD community (BMDO, DARPA, and ONR) have we seen the development of SiC wafers industry in the US able to produce single crystal substrates with lower defect density capable of sustaining device development programs [8]. Today, there are three suppliers of SiC wafers in US (CREE, Sterling and Litton Airtron), one in Europe and one in Japan with additional companies evaluating the market potential [8]. Within the last five years,

PAGE 18

5 Table 1-1 Physical properties of important semiconductors for high voltage power devices. Table 1-2 Normalized unipolar figures of merit of important semiconductors for high voltage power devices.

PAGE 19

6 commercial devices and applications have emerged. Specifically, SiC is used for blue and green LEDs, Northrop-Grumman makes Static Induction Transistors for internal applications, and CREE has announced their recent SiC Zero Recovery Rectifiers. Due to its exceptional switching speed, a high operational temperature and a high breakdown electric field, SiC is having a profound impact on the design, topology and circuits used in the power electronics especially where space weight, power density and thermal management issues dominate as in mobile platforms; on aircraft, on vehicles and especially on ships. There have been many funded programs carried out at universities, industry and government laboratories to commercialize SiC power switching devices. These programs are designed to address device processing and fabrication technology, device physics and design, packaging and applications. The most recent results in Schottky and PiN diodes demonstrated exceptional switching speeds with little or no ringing. When SiC diodes were packaged with Si switches to form a simple hybrid inverter circuit, nearly all the stored energy was eliminated due to the fast turn-off/turn-on SiC diodes [8]. At this time, US has a large but distribute program in SiC power device technology, sustained with DOD programs at universities and larger companies such as GE and Northrop Grumman. Traditional semiconductor solid-state switch manufacturers are now actively pursuing this technology. On the other hand, the more integrated European power industry, ABB and Siemens are actively pursuing research in SiC technology at all levels in an internally coordinated manner. In Asia, Japan has an active research program in SiC power technology that is now becoming focused under a new initiative [8].

PAGE 20

7 1.3 SiC Semiconductor Devices for Power Applications Silicon carbide offers significant advantages for power-switching devices because the critical field for avalanche breakdown is about ten times higher than in silicon. SiC power devices have made remarkable progress in the past five years, demonstrating currents in excess of 100 A and blocking voltages in excess of 19000 V [9]. Over the past decade a number of new SiC-based power switching devices have been demonstrated such as GTOs, MOSFETs, pn diodes and schottky diodes with results approaching or exceeding the theoretical limits of Si. However, there is significant difference between demonstrating prototype devices and establishing an economically viable commercial product. Silicon has proven to be an extremely difficult material to displace, largely because of economic factors, and it remains to be seen whether SiC can accomplish what other semiconductor materials have failed to do. There are significant new challenges emerging, both in material science and device design as the SiC technology is developed. One of the unique features of SiC is its high thermal stability compared to silicon. This has both advantages as well as disadvantagesan advantage because of improved reliability and higher temperature operability, and a disadvantage because many of the typical fabrication processes either do not occur at all in SiC or require extremely high temperatures. The most challenging aspect of the thermal stability is the fact that due to the phase equilibrium in the Si-C system, SiC does not melt, but instead gradually sublimes at temperatures above 2000 K. This makes it impossible to form large single-crystal ingots by pulling a seed crystal from a melt, as in Czochralski process that produces 200-300 mm diameter silicon ingots. Instead a modified sublimation process that is presently limited to 100 mm diameter ingots forms SiC crystals. The process is expensive and difficult and SiC wafer costs are astronomical

PAGE 21

8 compared to silicon. However, the advantages of SiC power devices are so great that the higher cost of the starting material can still be offset in certain cases, allowing SiC to compete directly with silicon in specialized device markets. The evolution of SiC power-switching technology can be divided into three phases: prototype device demonstration, device scale-up, and commercial production. Many types of SiC power devices are well along in the prototype demonstration phase and several have demonstrated performance measures far superior to silicon. These include rectifiers with blocking voltages above 19 kV [10], power metal-oxide-semiconductor field-effect transistors (MOSFETs) [11], and bipolar junction transistors (BJTs) [12] with performance figures 100 times higher than in silicon, and gate turn-off (GTO) thyristors blocking over 3 kV and switching over 60 kW. Several of these devices are in the scale-up phase, with terminal currents in excess of 100 A already demonstrated in single-device packaged rectifiers [13]. 1.3.1 Material Availability and Quality SiC has a large number of different crystallographic forms or poly-types. Of these, 6H-SiC has the most developed growth technology on account of its relatively large volume usage as a substrate for GaN blue LEDs. However, 4H-SiC is the preferred material for many power applications owing to its nearly ten times higher on-axis mobility compared to 6H-SiC. At present commercial 4H-SiC wafers are available in sizes up to 50 mm diameter while 6H-SiC wafers are available in sizes up to 75mm diameter, with 100 mm at the research stage. Substrates are available in both low resistivity (nand p-type) and semi-insulating forms [14]. However, the moderate to high resistivity substrates desirable for high voltage devices is not available. Devices are therefore, fabricated on homoepitaxial layers, which are routinely grown to thicknesses of

PAGE 22

9 over 100 m with doping densities as low as 1014 cm-3 using hot wall chemical vapor deposition (CVD) [15]. High-level carrier lifetimes for this material are typically of the order of several hundred nanoseconds [16] making it suitable for the fabrication of devices with voltage blocking capabilities approaching 10 kV. Historically, the major difficulty with SiC has been the presence of micropipes in the substrates and epilayers. It is generally conceded that micropipes are the major obstacles in the production of high-performance SiC devices. Micropipes are defects unique to the growth of SiC and they are physical holes that penetrate through the entire crystal. They are replicated into the device epitaxial layers and become killer defects if they intersect the active region of SiC devices. However, the reports on material growth with micropipe densities as low as 0.1/cm2 at research level (reduced from over 1000/cm2 in just a few years) [14], indications are that SiC is now adequate to fabricate devices several millimeters square with reasonable yield [17, 18]. Indications are that a dislocation density of less than 103 cm-2 is desirable for fabrication of power devices [18] (current values are typically around 104 cm-2). A further difficulty concerns the uniformity of epilayer doping and thickness across the wafer (typical results show 4 % standard deviation in doping) and the uniformity of doping between runs (typically 40%) [14]. 1.3.2 SiC Processing Technology SiC has a relatively mature processing technology. All the basic process steps, such as doping (by implantation and during epitaxy), etching (plasma techniques), oxide growth, Schottky and ohmic contacts have been successfully demonstrated [19]. There

PAGE 23

10 are, however, several key processing issues that currently limit the performance of fabricated devices. 1.3.2.1 Ion implantation Ion implantation is employed widely for local p-type and n-type doping of SiC. The most frequently used implanted ions are aluminum (Al) and boron (B) for p-type, nitrogen (N) and phosphorous (P) for n-type. Key issues in SiC implantation technology include the reduction of lattice damage occurring during implantation, successful electrical activation of the dopants and preservation of good surface morphology. Performing the implantation at elevated temperature is a common way to reduce the lattice damage and reduce the requirements for subsequent annealing [20]. Typical implantation temperatures are: 400 oC for Al [21], 500 oC for N [22], 800 oC for P [23]. These temperatures are sufficiently high to obtain good electrical conductivity after annealing but not so high as to cause surface dissociation and/or large vacancy clusters [24]. The annealing condition must be precisely determined, both for recording the crystal and diffusing the dopants into substantial sites. In addition, evaporation of both silicon and dopants must be avoided if the surface stoichiometry is to be preserved. This is particularly important in cases where the surface layer forms an active part of the device, for example in MOSFET and MESFET channel regions. There are two main approaches to preserve the surface: (i) material encapsulation with AlN, graphite [25]; (ii) increased Si partial pressure in the furnace reactor, either with silane gas [26] or with SiC coated pieces [27]. Several authors have given results for optimum electrical activation [28-30]. In all cases, it appears that good electrical activation is only possible if the annealing temperature exceeds 1550 oC.

PAGE 24

11 1.3.2.2 Oxides on SiC One of the principal benefits of silicon carbide is that it oxidizes to form a stable surface layer of silicon dioxide releasing carbon dioxide in the process [31]. However, the detailed properties of the oxide and in particular the interface between the SiC and the SiO2 are significantly different from Si [31]. The oxidation rate is crystal-orientation dependent and is far slower on the Si face than the C face, though, in general, much better properties are found for the oxide on the Si face. Oxidation temperature are normally around 1100 oC and unlike Si, a post oxidation anneal in a hydrogen ambient is usually reported to have little effect at reducing interface state density [32]. Interface state density on p-type material can be reduced to levels below 1011 cm-2eV-1 in the lower half of the gap by the use of re-oxidation anneal below 1000 oC in a wet ambient [33]. However, this treatment does not seem to improve the density of states in the upper half of the gap [34]. Interface state densities on the 4H polytype rise very rapidly towards the conduction band edge, typically exceeding 1013 cm-2eV-1, whereas on 6H the density is an order of magnitude lower. This has been ascribed to a carbon related acceptor that is located just below the conduction band edge for 4H but within the conduction band for 6H [35]. These defects have not been successfully removed by any standard surface treatment or post oxidation anneal. The best N channel inversion mode MOSFETs fabricated on 6H show electron mobilities of ~100 cm2/Vs with a negative temperature coefficient, whereas 4H shows mobilities of at best 25 cm2/Vs or lower [36, 37]. The fluctuation in potential resulting from the charge in the interface states near the conduction band edge appear to be responsible for these low values of mobility, particularly for 4H [38].

PAGE 25

12 The breakdown and reliability properties of SiO2 are crucial for all MOS devices, but unfortunately the situation is not as favorable as it is for Si. The dielectric constant of SiC is ~10 whereas it is ~3.9 for SiO2, so any normal surface field will be ~2.5 times higher in the oxide than the SiC. Hence to gain the full benefit of the 2.5 MV/cm breakdown field of SiC, the oxide must withstand 6.25 MV/cm, which is higher than is reliably usable, even on silicon [39]. High field stressing of oxides has shown that oxides grown in a wet ambient, breakdown at lower fields than dry grown oxides [40], and that extrapolated time-dependent dielectric-breakdown lifetimes of 10 years can only be obtained on n-type at fields less than 5 MV/cm at room temperature. Studies have shown that the lifetime for oxides drops rapidly at elevated temperatures, with a vulnerability to negative bias-stress instability [41]. Electron injection into the oxide is more efficient than for Si due to the lower barrier [32]. In addition, the barrier for injection of holes is much lower than in Si, which is unfortunate since holes are particularly damaging to oxides [42]. 1.3.2.3 Contacts Ohmic contacts to both n and p-type material have been demonstrated for 4H-SiC with specific contact resistivities of the order of 10-5 cm2 [43, 44]. To achieve this, it has proved necessary to use rapid thermal annealing (RTA) of the wafer at temperatures as high as 1400 oC. The majority of contacts are based around Ni and Al (for nand p-type, respectively) although these are generally not stable at temperatures above 400 oC. To take advantage of the full potential for high temperature operation of 4H-SiC devices, contacts that are thermally stable up to 600 oC over long periods are required. Recent work based on Al/Ni/W/Au contact structures on p-type 4H-SiC have shown longevity of over 100 h at 600 oC with a contact resistance of 10-3 cm2 [45]. Other works have

PAGE 26

13 shown similar results using a variety of materials including TaC [46] and AlSi [47] alloys. Several groups have reported excellent quality room temperature Schottky diodes, with the barrier height dependent on the metal chosen. This demonstrates that the fermi level is not pinned at the surface of the 4H-SiC. Values of B vary from 1.10 eV (Ti) to 1.73 eV (Au), which are markedly larger than those observed on Si or GaAs. This increase in B makes SiC an ideal material for high-voltage, high-temperature, low leakage current Schottky diodes. Surface preparation, prior to deposition of the Schottky contact, has been shown to be a key factor in achieving good performance with sacrificial thermal oxidation being the optimum technique [48]. 1.3.2.4 Edge termination and passivation Many techniques applied to Si devices are also applicable to SiC. For example, field plates [49, 50], guard rings and junction termination extensions [51, 52] have all been used to good effect. Another simple technique involves the implantation of a high dose of inert ions (either Ar [53] or B [54]). In this case the damage caused by the implant causes a high resistivity region to be formed close to the surface, which acts in a manner similar to semi-insulating poly crystalline silicon (SIPOS). The reverse leakage performance of Ar and B implanted edge terminations may be improved by low temperature (600 oC) annealing [55]. Passivation of SiC surface is not trivial and has yet to be fully understood. The effects of inadequate passivation have been observed in microwave MESFETs, where this leads to degradation of gain under CW operation [56]. Power switching devices, on the other hand, appear to perform well with conventional Si passivation treatments such as polyamide.

PAGE 27

CHAPTER 2 SILICON CARBIDE SCHOTTKY RECTIFIERS Silicon carbide device technology has seen a tremendous development over last five years. The technology once envisioned as a substitute to silicon technology has been realized to practice. The feasibility of SiC devices has been shown for many different types of devices, the development of a working production technology has started, yield, reliability and costs now being the key issues. The high cost of SiC substrates at present poses a major challenge for the technology to enter the device market. However, the exceptional properties of the material promote a tremendous interest in some of the prime applications like power electronics and high temperature sensor technology. The other examples are in transportation and manufacturing sector where high efficiency needs entails strong research efforts towards development of advanced power management and control electronics. Central to this effort is the development of the solid-state devices capable of delivering large currents and withstanding high voltages, without the need of sophisticated cooling systems [56]. 2.1 Experimental Characteristics of 4H-SiC Schottky Rectifiers 2.1.1 Introduction SiC has shown outstanding potential for high power, high temperature electronics in advanced communications and radar systems, sensor and control systems, utility power switching and traction motor control [57-68]. The wide bandgap (3.25 eV for the 4H-poly type), availability of doped substrates of either electrical conductivity type and high thermal conductivity (up to 4.9 W/cmK) of SiC make it a preferred material for these 14

PAGE 28

15 applications [69-77]. These characteristics result in significantly higher current densities, operating temperatures and breakdown voltages than comparable Si devices and in lower switching and on-state losses. There have been major advances in the bulk and epitaxial growth of SiC in the past 5 years or so, along with improvements in device design and fabrication. Those have resulted in very impressive performance from SiC power metal-oxide semiconductor field effect transistors (MOSFETs), bipolar transistors and p-i-n and Schottky rectifiers [78-93]. 2.1.2 Experimental Methods The starting substrates were n+ (1019 cm-3) 4H-SiC. Approximately 10m of lightly n-type (n ~ 5 x 1015 cm-3) 4H-SiC was grown on these substrate by vapor phase epitaxy, followed by ~ 500 of thermal SiO2. A full-area back contact of e-beam deposited Ni annealed at 970 0C was used for contact to the substrate. SiO2 layers deposited by plasma enhanced chemical vapor deposition (PECVD) were employed as a part of the dielectric overlap edge termination [94]. Holes were opened in the SiO2 stack by a combination of dry and wet etching and a Schottky contact of e-beam evaporated (1000 thick) Ni patterned by lift-off. The contact diameter ranged from 100-1000 m. A schematic of the completed structure and an optical micrograph of a processed 2 inch diameter wafer are shown in Figure 2-1. A key feature of achieving good Schottky performance is the quality of the ohmic contacts. The contact resistivity ~1.2 x 10-6 cm2 was obtained from transmission line measurements of the Ni backside ohmic for 970 0C, 3 min anneals under a flowing N2 ambient in a Heatpulse 610 furnace. This is slightly higher than reported for annealed Ni

PAGE 29

16 PECVD SiN x Thermal SiO 2 Epi nSiC (t epi =10 m ) Substrate n+ 4H/SiC ( n+= 1e19 ) Ni Ni Figure 2-1 Cross-section of SiC Schottky rectifiers (top) and photograph of completed 2 inch diameter wafer (bottom).

PAGE 30

17 on 6H-SiC substrates (7 x 10-7 cm2) [95], which have a slightly smaller bandgap (3.02 eV) than the 4H-polytype used here. 2.1.3 Results and Discussion Figure 2-2 shows some reverse current-voltage (I-V) characteristics measured at 25oC schottky contact diameter. Note that the reverse breakdown voltage decreases significantly with increasing contact size. This is due to the increased probability of having crystal defects in the active region of the device as the area is increased. The production of SiC substrates with large, defect free areas still represents the biggest challenge to widespread use of power rectifiers in this materials system. Note also the very low reverse leakage currents in the rectifiers at biases below breakdown. In this bias regime, the reverse current was proportional to the length of the rectifying contact perimeter, suggesting surface contributions were the most important in this voltage range. At biases closer to breakdown, the current was proportional to the area of the rectifying contact. Under these conditions, the main contribution to the reverse current is from rectifiers with a 7000 thick SiO2 dielectric overlap layer, as a function of the under this contact, i.e. from the bulk of the material. A more detailed view of the reverse I-V characteristic from a 154 m diameter rectifier is shown in Figure 2-3 (top), along with a map of VB values measured over a quarter of a 2-inch diameter wafer (bottom). The overall shape of the I-V curve is similar to past reports for Ni/4H-SiC rectifiers [90], in which it was concluded that the origin of the reverse current was due to a combination of both thermionic field emission and field emission. Note that yield of rectifiers with VB > 750 V is over 50 % and with VB > 600 V is over 75 %.

PAGE 31

18 -800-600-400-200010-1210-1010-810-610-4 100 m diode 500 m diode 1000 m diode Current (A)Bias Voltage (V) Figure 2-2 Reverse I-V characteristics at 25 0C from Schottky rectifiers with different contact diameters.

PAGE 32

19 -1000-800-600-400-200010-1110-1010-910-810-710-610-5 SiC Diode 5000 SiO2Dia: 154 m Current (A)Bias Voltage (V) 520 V 600 V 860 V 980 V 800 V 900 V 520V 940 V 860 V 820 V > 1000 V 830 V 660 V 640 V 750 V 810 V 810V > 1000 V 930 V 580 V 400 V 560 V 820 V >1000 V 720 V 740 V 300 V > 1000 V 820 V 750 V 600 V 290 V 980 V 730 V dam a ged 500 V 720 V Figure 2-3 Reverse I-V characteristic from 154 m diameter rectifier (top) and map of reverse breakdown voltage from a quarter of a 2 inch diameter wafer (bottom).

PAGE 33

20 Simulations of the rectifier performance were carried out using the MEDICITM code. The performance of the structure of Figure 2-1 (top) was simulated for different oxide thicknesses and a metal overlap of 10 m onto this oxide, as used experimentally. Figure 2-4 shows a comparison of the experimental and theoretical results. Note that the VB values obtained on the real devices are roughly half of that expected from the simulations. The later do not take into account the presence of crystalline defects, including micropipes in the SiC. This difference emphasizes the importance of crystal quality in determining the performance of the rectifiers. The reverse breakdown voltage is related to the epilayer doping (ND) through the relationship [96]: N EVDCBe22 where is the dielectric constant of SiC, EC the critical field for breakdown and e the electronic charge. Figure 2-5 shows our experimental data point for VB at a epilayer doping concentration of 5 x 1015 cm-3 for 10 m thick layers, along with the expected results from the simulations with different carrier concentrations. Note that there is not much improvement expected for decreasing the carrier concentration below 5 x 1015 cm-3 since the layer will be depleted under those conditions. Our experimental devices still have performance limited by material defect density rather than purity. The VB values obtained in the simulations were found to increase linearly with SiO2 thickness from 2500 to 17500 and saturated thereafter.

PAGE 34

21 200030004000500060007000800090001000011000-1500-1400-1300-1200-1100-1000-900-800-700 SiC Schottky DiodesDoping n:5e15 cm-3tepi:10 m Experimental results Simulated results Breakdown Voltage (V)Dielectric thickness () Figure 2-4 Comparison of breakdown voltage obtained from simulations and the fabricated devices.

PAGE 35

22 However, thicknesses above ~5000 are difficult to achieve in practice because of the long deposition times. The forward current characteristics of a Schottky rectifier are dominated by the barrier height of the metal on the semiconductor and by series resistance concentrations. The barrier height (B) for Ni on 4H-SiC is ~1.3eV [90]. The forward voltage drop, VF, is related to the barrier height and on-state resistance, RON, through the relation: JR T A JVFONBFFnenkT.)ln(2* where n is the diode ideality factor, k is Boltzmanns constant, T the absolute temperature, e the electric charge, JF the forward current density and A* is Richardsons constant. The value of RON is a function of epilayer thickness and doping. Figure 2-6 shows forward I-V characteristics in both linear (top) and log (bottom) form. A maximum forward current of 2 A was obtained from these devices, and was limited by our experimental system. The turn-on voltage was ~1.82 V at a forward current density of 100 Acm-2 and the on-state resistance was ~0.36 mcm2. The diode ideality was 1.13 at 25 0C, which is comparable to past results for Ni metallization [90]. To place our results in context, Figure 2-7 shows a theoretical plot of specific on-resistance as a function of VB for Si, 4H-SiC and 6H-SiC, along with some experimental data. The lines are derived from the relation [84-96]: E VRCBON324

PAGE 36

23 1014101510161017102103 Simulation Results Experimental ResultSiC Schottky Diode tepi=10 mtdielectric=1.0 m Breakdown Voltage (V)Carrier Concentration (cm-3) Figure 2-5 Calculated reverse breakdown voltage as a function of epilayer doping.

PAGE 37

24 0.00.51.01.52.02.53.03.50.00.51.01.52.02.5 SiC Schottky DiodesSiO2: 7000 Diode Dia.: 1000 m Current (A)Bias Voltage (V)0.00.51.01.52.02.53.0100101102 SiC Schottky DiodesSiO2: 7000 Diode Dia.: 1000 m Current Density (A/cm2)Bias Voltage (V) Figure 2-6 Forward I-V characteristics in linear (top) and log (bottom) scales.

PAGE 38

25 Figure 2-7 Specific on-resistance as a function of breakdown voltage.

PAGE 39

26 where is the electron mobility in the epilayer. Note that the SiC rectifiers are expected to have either much lower on-resistance at a given breakdown voltage than Si, or equivalently a much larger breakdown voltage for a given on-resistance. Our results are close to the theoretical minimum on-resistance for the VBs obtained and indicate that the device processing produces no major degradation in the electrical properties of the SiC. 2.2 Ni Contacts to N-Type 4H-SiC Superior electrical and thermal properties of SiC make it promising material for high temperature and high frequency electronics. However, it is widely appreciated that the optimal performance of microwave devices is related to the quality of ohmic contacts. Low contact resistance and high temperature stability are required [97]. Nickel is one of the most popular materials used in advanced aerospace systems [98] and good nickel ohmic contacts on n-SiC have been already demonstrated [99]. The popularity of Ni contacts to SiC is due to its reproducibly low contact resistance and good temperature stability [100]. In our study, we could see no appreciable degradation of the contacts up to a temperature of 225 oC. Annealing the contacts produces low ohmic contact resistivities on highly doped substrates. For n-type SiC, low contact resistivities (<10-5 cm-2) have been reported in the past [101]. We obtained a contact resistivity of 1.2 x 10-6 cm2, in our studies. A study on the interfacial reactions between Ni and n-type SiC suggests that silicides such as Ni31Si12 and Ni2Si are produced at an intermediate temperature range of 500-600 oC [102]. X-ray diffraction (XRD) and Rutherford backscattering spectroscopy (RBS) showed that the interfacial reaction forming Ni31Si12 and Ni2Si began at 500 oC [103], and the Ni2Si phase remained up to 900 oC. The ohmic contact formation was thought to be

PAGE 40

27 due to the formation of Ni2Si phase, even the Ni2Si formed at a temperature as low as 600 oC [101]. It is essential to optimize the annealing conditions to obtain contacts with least contact resistivity. In our studies we have studied the effect of temperature, ambience and time on the contact resistivity of Ni-SiC ohmic contacts. Figure 2-8 shows the effect of annealing temperature (top-left), time (top-right) and annealing ambient (bottom) on the contact resistivity obtained from transmission line measurements of the Ni backside ohmic. A minimum value of ~1.2 x 10-6 cm2 was achieved for 970 0C, 3 min anneals under a flowing N2 ambient in a Heatpulse 610 furnace. This is slightly higher than reported for annealed Ni on 6H-SiC substrates (7 x 10-7 cm2) [95], which have a slightly smaller bandgap (3.02 eV) than the 4H-polytype used here. 2.3 Stability of SiC Schottky Rectifiers to Rapid Thermal Annealing 2.3.1. Introduction There is currently a lot of interest in the application of SiC power rectifiers to high current switching in the >30A range because of the high thermal conductively (~4.9W/cmK) and large bandgap relative to Si [104-112]. The advantage in materials properties mean that on-state resistance, RON, for a 4H-SiC rectifier can be over 100 times smaller than that of a Si rectifier at the same breakdown voltage, VB, since RON is given by 4VB/EC3 (where is the electron mobility, the permittivity, and EC the critical field for breakdown) [87]. There are numerous situations in which high temperature annealing might be employed in the fabrication of SiC rectifiers, including activation of implanted dopants or reduction of reverse leakage current in self-aligned Ar+-implanted edge terminations

PAGE 41

28 95096097098099010001.1x10-61.5x10-61.9x10-62.3x10-6 Anneal time: 3 minAnneal gas: N2 Rc (Ohm-cm2)Temperature (C)2.02.53.03.54.01.0x10-61.4x10-61.8x10-62.2x10-6 Anneal temperature: 970 CAnneal gas: N2 Rc (Ohm-cm2) Time (min)N2VacuumAr1.1x10-61.3x10-61.5x10-61.7x10-6 Anneal time:3 minAnneal temp:970 C Rc (Ohm-cm2)Anneal Gas Figure 2-8 Contact resistivity of Ni ohmic contact on n+ 4H-SiC as a function of annealing temperature (top), annealing time (center) and annealing ambient (bottom).

PAGE 42

29 [113-115]. In the latter case, Ar+ implantation is used to create a high-resistance region around the contact periphery in order to spread the electric field distribution and avoid breakdown due to field crowding. While this process does increase reverse breakdown voltage, it also increases leakage current. Several studies have examined the effect of anneals up to 700C for reducing this leakage current [114]. In this study we report the stability of 4H-SiC Schottky power rectifiers to rapid thermal annealing treatments in the temperature range 700 1100C with the contacts already in place. This is of interest for defining the stability of devices in which the Ni rectifying contact is used as a self-aligned mask for implant edge termination. The performance of the rectifiers is found to show a general improvement for anneals 1000C. 2.3.2 Experimental Methods The Schottky contacts were fabricated on the wafer as described in section 2.1.2. The annealing for these contacts was performed in a Heatpulse 610T System under flowing N2 ambient at temperatures of 700 1100 C for 60 240 sec. The current-voltage (I-V) characteristics were measured before and after this procedure. Prior to annealing, the typical on-state resistance was 0.8 mcm2, the reverse breakdown voltage (VB) ~450V and the forward turn-on voltage was 2.5V at 100Acm-2. 2.3.3 Results and Discussion The reverse bias characteristics showed a general increase in current for anneals 1000C, while the forward I-V characteristics also shoed significant changes under these conditions, as shown in Figure 2-9. Note the decrease in turn-on voltage as the anneal temperature increase. The forward current density, JF, can be expressed as:

PAGE 43

30 1exp2**nkTeVkTeTAJBF where A** is the Richardsons constant for SiC, k is Boltzmanns constant, e is the electric charge, B is the barrier height for Ni on 4H-SiC, V is the applied voltage, n is the ideality factor, and T is the measurement temperature. The data in Figure 2-9 are consistent with a reduction in B. On 6H-SiC, Ni was observed to form Ni2Si at temperature beginning at ~600C [116], eventually leading to ohmic behavior. In lower temperature anneals, Ni/6H-SiC diodes showed a reduction of leakage current through removal of low-barrier secondary diodes in parallel to the primary diode [113]. Our measure B was ~1.4eV prior to annealing, which is consistent with past reports [116-118]. It is likely that the reduced turn-on voltage observed in Figure 2-9 is the result of silicide formation, as suggested previously [114]. The on/off current ratio was 1.5 105 at 3V/-450V in control samples and increased by a maximum of ~20% after optimum annealing at 700C. The forward current characteristics of a SiC Schottky rectifier are dominated by the Ni barrier height and by series resistance contributions and is related to B and RON through the relationship: FONBFFJRnTAJenkTV2**ln With annealing, the value of B decreased as described above, but RON showed improvement (Figure 2-10). Once again the annealing durations had to be kept<120 sec at the highest temperature in order to retain some acceptable rectifying behavior of the top Ni contact.

PAGE 44

31 Figure 2-11 shows the effect of annealing time at 900C on the forward I-V characteristics (top), VF (center), and RON (bottom). Note that the largest anneal (240 sec) produced essentially ohmic behavior. Examples of the effect of the annealing duration on the top Ni contact morphology are shown in the optical micrograph of Figure 2-12. It is clear that there is already a strong metallurgical reaction between the Ni and the SiC at 1000C for 60 sec. To obtain the lowest ohmic contact resistivity for annealed Ni contacts requires ~3 min at this temperature, and after 60 sec the contact still retains some rectifying behavior. Note also the dark appearance of the contact after annealing. Auger Electron Spectroscopy showed high levels of carbon present in these reacted contacts, suggesting the reaction can be written as: 2Ni + SiC Ni2Si + C Figure 2-13 shows the measurement temperature dependence of the change in VF (top) and RON (bottom) for samples annealed at 900C. There is little significant change in VF while the improvement in RON decrease with increasing temperature as current transport over the barrier. Arrhenius plots of the forward current indicated a barrier height of ~1.14 eV after the 900C anneal, consistent with the observed increase in current after annealing. The intersection of the lines with the temperature axis indicated a constant electrically active contact area equivalent to ~9% of the physical contact area. This is consistent with the expected behavior for thermionic emission still being the dominant transport mechanism after 900C annealing.

PAGE 45

32 2.3.4 Summary and Conclusions The main points of our study may be summarized as follows: High voltage Ni/4H-SiC Schottky rectifiers using a simple SiO2/metal edge termination method show near-ideal forward characteristics (ideality factor, on-state resistance). The reverse breakdown voltage is approximately half that expected from a simple model simulation and decreases as the device area is increased. The reverse current appears to be dominated by surface combinations at low bias and by bulk combinations near breakdown. Both thermionic field emission and field emission mechanisms appear to be present. The yield of 750 V, 2 A devices was above 50 %. For high temperature anneals, the Ni rectifying contact becomes ohmic, with severe degradation of the contact morphology. The forward current characteristics are still dominated by thermionic emission until the onset of ohmic behavior. The benefits of annealing at moderate temperature include a decrease in forward turn-on voltage and an increase in the on/off current ratio. The reasons for those improvements may include annealing of defects or reduction of interfacial oxides between the Ni rectifying contact and the SiC. The use of more thermally stable rectifying contact metallization would allow higher annealing temperatures and possibly more improvement in device performance.

PAGE 46

33 0.00.51.01.52.02.53.00.0000.0050.0100.0150.0200.0250.030 60 secanneals1000 C900 CControl Current (A)Bias Voltage (V) Figure 2-9 Forward I-V characteristics as a function of anneal temperature for 60 sec anneals.

PAGE 47

34 7007508008509009501000105011000.00.51.01.52.02.53.03.54.0 60 secanneals RON (10-3)Temperature (C) Figure 2-10 Change in RON as a function of anneal temperature for 60 sec anneals.

PAGE 48

35 0.00.51.01.52.02.53.03.54.00.0000.0050.0100.0150.0200.0250.0300.035 240 secs60 secsControl Current (A)Bias Voltage (V)406080100120140160180200220240260-2.0-1.9-1.8-1.7-1.6-1.5-1.4-1.3-1.2-1.1 VF Anneal Time (Seconds)4060801001201401601802002202402600.000.250.500.751.001.251.501.752.002.25 RON(10-3)Anneal Time (Seconds) Figure 2-11 Forward I-V characteristics (top) and change in VF (center) and RON (bottom) as a function of anneal time at 900C.

PAGE 49

36 Figure 2-12 Optical micrographs of Ni contacts before (top) and after annealing at 1000C for 60 sec (center) or 120 sec (bottom).

PAGE 50

37 20406080100120140160-0.55-0.50-0.45-0.40-0.35-0.30-0.25-0.20 VFMeasurement Temperature (C)204060801001201401600.20.40.60.81.01.21.41.61.82.0 RON(10-3)Measurement Temperature (C) Figure 2-13 Change in VF (top) and RON (bottom) as a function of measurement temperature after 900C, 60 sec anneals.

PAGE 51

CHAPTER 3 IRRADIATION AND PASSIVATION EFFECTS ON 4H-SIC RECTIFERS 3.1 High Energy Proton Irradiation Effects on SiC Schottky Rectifiers 3.1.1 Introduction SiC power rectifiers with reverse breakdown voltages in the 300-1200 V range are of current interest for a number of applications, including traction motor control, sensors and advanced communications (including broad-band satellite transmission) and radar systems [105-108]. In particular, 4H-SiC is attractive because of its larger bandgap and higher and more anisotropic mobility relative to other SiC polytypes [119]. The 4H-SiC has numerous advantages over Si for power device applications, including much larger critical breakdown field, lower on-state resistance at a given voltage and higher thermal conductivity. While a significant amount of work has been reported on ion-beam induced damage and annealing in SiC [120-123], much less is known about radiation damage in SiC-based devices. Recent results have shown that properly passivated SiC metal-oxide semiconductor field effect transistors have good tolerance to MRad doses of 60Co -rays [124]. One would expect SiC to be relatively resistant to radiation damage compared to materials such as GaAs based on the empirical observation that the displacement threshold energy for atoms in semiconductor is inversely proportional to lattice constant [125] (a = 0.31 nm, c = 1.03 nm for 4H-SiC compared to a = 0.5653 nm for the zinc blende GaAs). There is particular interest in the response of SiC rectifiers to high-energy proton irradiation because of the potential applications in satellites. In this work we reported on 38

PAGE 52

39 the effects of 40 MeV proton irradiation of 4H-SiC rectifiers, at doses corresponding to more than 10 years in low-earth orbit. The reverse leakage current increases by approximately a factor of two under these conditions, while ideality factor, on-state resistance and forward turn-on voltage all increase by 75%. These results are consistent with creation of electron traps by protons that traverse the active region of the rectifiers. 3.1.2 Experimental Methods The Schottky rectifiers with diameter of 154 m were fabricated in the same way as described in section 2.1.2. The devices were irradiated at the Texas A&M cyclotron with 40 MeV protons at fluences from 5 107-5 109 cm-2, with the highest fluence being comparable to at least 10 years exposure in low earth orbit. The projected range of the protons is >50 m in SiC. The irradiated devices were measured approximately 2 days after exposure. 3.1.3 Results and Discussion Figure 3-1 (top) shows reverse current voltage (I-V) characteristics before and after a proton fluence of 5 109 cm-2. The basic shape of the characteristic is unchanged and shows only an increase in the magnitude of the reverse leakage current. Figure 3-1 (bottom) shows the percentage increase in reverse current measured at V, as a function of the proton fluence. We suggest the creation of generation-recombination centers related to atomic displacements created by the incoming protons is the main mechanism for the increased leakage current, based on both this data and what follows later in this section. The forward I-V characteristics before and after proton irradiation at fluence of 5 x 109 cm-2 are shown at the top of Figure 3-2. There is an increase in low-bias (< 1V)

PAGE 53

40 current consistent with the introduction of recombination centers associated with proton-induced displacement damage. At higher forward voltages, the current is decreased. Figure 3-2 (bottom) shows the magnitude of the decrease (measured at a forward voltage of 2 V) as a function of the proton fluence. Note the percentage decrease in current is ~ 6% after a fluence corresponding to more than a year in low earth orbit and 42% after more than 10 years. The Schottky barrier height was extracted from the relationship: ]1))[(exp(2**nkTeVkTeBFTAJ where JF is the forward current density at voltage V, A** is the Richardsons constant for 4H-SiC, T the absolute measurement temperature, e the electronic charge, B the barrier height for Ni on 4H-SiC, n the ideality factor and k the Boltzmanns constant. The extracted values were 1.32 0.06 eV for both control and irradiated rectifiers, which is consistent with past reports for Ni on 4H-SiC [54]. The forward voltage drop for a Schottky rectifier, VF, is related to the barrier height and on-state resistance, RON, through: JR T A JVFONFFnenkT)ln(2** VF is usually defined as the voltage at which JF is 100 A.cm-2. Figure 3-3 shows the increases in VF, n and RON as a function of proton fluence. The values in the unirradiated control rectifiers were VF = 2.4V, n = 1.15 and RON = 3.7 10-4 cm2. In each case, the values of these parameters increase with increasing proton fluence. The results are consistent with a net reduction in carrier density in the depletion region of the rectifiers through the introduction of traps and recombination centers associated with proton damage.

PAGE 54

41 3.2 Influence of PECVD of SiO2 Passivation Layers on 4H-SiC Schottky Rectifiers 3.2.1 Introduction There is great current interest in the development of manufacturable SiC power rectifiers in the 300-1000 V, 25-50 A range for use in traction motor control, sensor and control systems and the drive train of hybrid electrical vehicles [119-126]. Numerous reports have shown the potential of both SiC Schottky and p-i-n rectifiers for these applications [84, 87]. Record breakdown voltage (12.3 kV) and forward current (130 A) results have been achieved for these devices [87]. The main focus now is to understand the effect of crystal growth and device processing steps in the performance of SiC rectifiers, in an attempt to push the devices into a robust manufacturing mode. One of the important fabrication steps is the deposition of the thick dielectric layers for use as surface passivation and as a part of the metal overlap edge termination. These layers are generally deposited by plasma enhanced chemical vapor deposition (PECVD) using SiH4-based chemistries and therefore the surface of the SiC is subject to both energetic ion bombardment and a high flux of atomic hydrogen. The former can create surface and bulk traps that degrade the electrical properties of semiconductors while the later can etch the surface through the reaction: SiC + 8H SiH4 + CH4 as well as passivate dopants in the near-surface region. Both donors (D) and acceptors (A) can be passivated through the formation of neutral complexes with hydrogen, according to following mechanisms: D+ + H(DH)0 A+ H+ (DH)0

PAGE 55

42 -500-400-300-200-100010-1110-1010-910-810-710-610-510-4 Post-irradiation Pre-irradiationdose: 5x109 p/cm2SiC Schottky diodeSiO2:7000 Dia.: 1000 m Current (A)Bias Voltage (V)1071081091010100101102103 SiC Schottky diodeSiO2:7000 Dia.: 1000 m % increase in IL (-250V)Proton Dose (cm-2) Figure 3-1 Reverse I-V characteristics from 4H-SiC Schottky rectifiers before and after proton irradiation at a dose of 5 x 109 cm-2 (top) and percentage increase in reverse leakage at V (bottom).

PAGE 56

43 0.00.51.01.52.02.53.010-210-1100 dose: 5x109 p/cm2 pre-irradiated post-irradiated SiC Schottky diodeSiO2:7000 Dia.: 1000 m Current (A)Bias Voltage (V)107108109101001020304050 SiC Schottky diodeSiO2:7000 Dia.: 1000 m % decrease in IF (2V)Proton Dose (cm-2) Figure 3-2 Forward I-V characteristics from 4H-SiC Schottky rectifiers before and after proton irradiation at a dose of 5 x 109 cm-2 (top) and percentage decrease in forward current at 2V (bottom).

PAGE 57

44 10710810910100.00.51.0 n Proton Dose (cm-2)107108109101001234 RONx 10-4 cm2Proton Dose (cm-2)10710810910100.00.51.0 VF (V)Proton Dose (cm-2) Figure 3-3 Numerical change in values of n (top), RON (center) and VF (bottom) as a function of proton fluence.

PAGE 58

45 In this study we report on the effects of PECVD SiO2 layers on the performance of Ni/4H-SiC Schottky rectifiers. We also identify optimum deposition conditions under which there is minimal degradation of the device performance. 3.2.2 Experimental Methods The Schottky rectifiers with a diameter of 154 m were fabricated in the same way as described in section 2.1.2. PECVD of thin (~200 ) SiO2 films were performed at a substrate temperature of 250 0C using 2% SiH4 and N2O at a total flow rate of 600 sccm. The process pressure was raised from 600-900 mTorr, the N2O percentage of the total flow rate from 5 to 50% and the rf power (13.56 MHz) varied from 25-400W. The SiO2 layers were thin enough that we could probe through them to the underlying contacts, allowing us to measure the current voltage (I-V) characteristics without having to remove the SiO2 layers. Prior to SiO2 deposition, the rectifiers displayed reverse breakdown voltage (VB) of 770 V, on-state resistance (RON) of 4.73 m.cm2 and forward turn-on voltage (VF) of 2.06 V at 100 A.cm-2. 3.2.3 Results and Discussion Figure 3-4 shows the effect on VB of plasma power (top), pressure (center) and N2O percentage (bottom) during the SiO2 deposition. There is no significant change in VB at low powers, while at higher powers (>250 W), it increases. Since VB is inversely dependent on the doping in the depletion region (ND), according to the relation: N EVDCBe22 where is the permittivity of SiC and e the electronic charge, this suggests a decrease in doping in the unprotected region around the rectifying contact (and possibly under, since atomic hydrogen ion diffuse in an isotropic fashion). This effect was largest at the lowest

PAGE 59

46 process pressure, as seen in the center part of Figure 3-4. The bottom part of the figure shows that VB actually decreased under SiH4-rich deposition conditions. Therefore, we suggest that the main effect of the hydrogen flux is to decrease VB through creation of surface states, whereas the effects of plasma power and pressure are mainly due to creation of deep traps by energetic ion bombardment. In this later case, the effective doping around the contact is decreased and this leads to an increase in breakdown voltage. The forward turn-on voltage was less sensitive to changes in the SiO2 deposition parameters than was VB. This is expected, since VF is controlled by the barrier height of Ni contact (1.3 eV) and by series resistance contributions, especially from contact resistance. Since the area under the Schottky contact is protected by the overlying metal and the rear ohmic contact is not exposed to the plasma, we would expect VF to be less sensitive to the process conditions. Figure 3-5 shows the effect on VF of both plasma power (top) and process pressure (bottom), with the changes in each case being less than ~ 10%. The on-state resistance is related to the critical field for breakdown EC and the reverse breakdown voltage through the relation [96]: E VRCBON324 where is the electron mobility in the 4H-SiC epi-layer. Figure 3-6 shows the effect of plasma power (top), N2O percentage (center) and process pressure (bottom) on RON. The trends mirror those seen for VB under the same conditions, with the increases being greatest at the highest powers, highest SiH4 content in the plasma and the lowest process pressure.

PAGE 60

47 3.3 Effect on 4H-SiC Schottky Rectifier of Ar Discharges Generated in a Planar Inductively Coupled Plasma Source 3.3.1 Introduction SiC-based rectifiers are generating tremendous current interest for a broad range of applications, including broad band satellite transmission systems, advanced radar, high temperature sensors and traction motor control [119-127]. Most attention has been focused on the 4H-SiC polytype because of its larger bandgap (3.25 eV) and higher mobility relative to the other polytypes [70]. Within the two basic classes of rectifiers, Schottky devices have the lowest on-state voltages and highest switching speeds; while p-i-n diodes have the higher reverse breakdown voltage and lower reverse leakage current [87]. Compared to Si rectifiers, SiC devices have on-state resistances approximately a hundred times lower or equivalently much larger breakdown voltage at the same on-state resistance [96]. While very high forward currents (up to 130 A) and breakdown voltages (4.9 kV) have been achieved for 4H-SiC Schottky rectifiers [87], a lot of recent attention has been focused on fabrication and materials technology for diodes in the 300-1000 V range [105]. Plasma processing is required both for dry etching of mesas and deposition of dielectrics for surface passivation and metal overlap edge termination. However little has been reported on the effects of plasma exposure on the electrical performance of 4H-SiC rectifiers. In this study, we describe the effect of Inductively Coupled Plasma (ICP) Ar discharges generated in a novel plasma source configuration, on the electrical properties

PAGE 61

48 0100200300400-30-20-1001020304050 SiC Schottky diodes % change in VbPower (W)600700800900-2002040 SiC Schottky diodes % change in VbPressure (mTorr)01020304050-100-80-60-40-200 SiC Schottky diodes % change in Vb% N2O Figure 3-4 Percentage change in VB as a function of plasma power (top), process pressure (center) and N2O content (bottom) during SiO2 deposition.

PAGE 62

49 010020030040024681012 SiC Schottky diodes % increase in VfPower (W)600700800900246810 SiC Schottky diodes % increase in VfPressure (mTorr) Figure 3-5 Percentage change in VF as a function of plasma power (top) and process pressure (bottom) during SiO2 deposition.

PAGE 63

50 01002003004000510152025 SiC Schottky diodes % increase in RonPower (W)010203040500246810 SiC Schottky diodes % increase in Ron% N2O6007008009000510152025 SiC Schottky diodes % increase in RonPressure (mTorr) Figure 3-6 Percentage change in RON as a function of plasma power (top), N2O content (center) and process pressure (bottom) during SiO2 deposition.

PAGE 64

51 of Ni/4H-SiC Schottky rectifiers. Both ion flux and ion energy are found to play a role in the extent of the observed changes in reverse breakdown voltage (VB), on-state resistance (RON), ideality factor (n) and forward turn-on voltage (VF). 3.3.2 Experimental Methods The Schottky rectifiers with a contact diameter of 154 m were fabricated in the same way as described in section 2.1.2. The as-fabricated rectifiers show RON of 4.4 m-cm2, VF = 1.96 V and VB = -330 V. The devices were exposed to pure Ar discharges in a planar coil geometry ICP reactor, shown schematically in Figure 3-7. A key feature of this system is the ability to operate at very low source power (20 W) as well as chuck power (5 W), which allows one to design etch or deposition processes that end with very low ion energy and flux conditions to minimize damage. The ICP source power (2 MHz) was varied from 100-700 W, the rf chuck power (13.56 MHz) from 25-200 W and the process pressure from 10-40 mTorr. All exposures were 30 sec in duration and removed < 150 of the surface by sputtering. These plasma exposure simulate the ion bombardment received by the SiC surface during processes such as dry etching of plasma enhanced chemical vapor deposition and represent a worst-case scenario because there is no chemical etch component that would remove some of the damaged surface region on a deposition of dielectric to protect the surface from further ion bombardment. 3.3.3 Results and Discussion Figure 3-8 shows the percentage changes in VB (top) and RON (bottom) as a function of ICP source power (100 W) and process pressure (10 mTorr). The reverse breakdown voltage decreases with increasing ion flux. Since the dc self-bias on the

PAGE 65

52 sample electrode also decreases with increasing source power, this indicates that increasing ion flux is a factor in degrading the breakdown voltage. Note however that the changes are 25% even at the highest source power employed. For SiC there is an empirical relationship between VB and the doping in the epilayer, ND, which is given by [96]: N V DB75.0151075.1 One of the expected effects of Ar plasma exposure on the exposed SiC around the contact periphery is a decrease in ND through creation of deep trap states. However this would lead to an increase in VB, which is the opposite of what is observed experimentally. Therefore, we suggest that the main effect of the plasma exposure under our conditions is an increase in surface defects that initiate breakdown at much lower values than expected from the bulk doping of the SiC epilayer. The on-state resistance can be written as [106]: E VRCBON324 where is the electron mobility, the permittivity of 4H-SiC and EC the critical field for breakdown. Since VB decreases with increasing source power, we would also expect a decrease in RON unless EC is also decreasing. Figure 3-8 (bottom) shows large increases in RON as the source power is increased, which is consistent with the creation of surface states that promote early breakdown of the damaged rectifier. Figure 3-9 shows the percentage changes in VF (top, measured at 100 A.cm-2) and ideality factor (bottom) as a function of source power at fixed rf power (100 W) and pressure (100 mTorr). Note that the changes in both parameters are 20% provided the

PAGE 66

53 source power is left below 300 W, corresponding to a dc self-bias -185 V. At higher source powers both VF and n are severely degraded even though the self-bias under these conditions is low (-41 V at 700 W source power). The average energy of the incident Ar+ ions is roughly given by the sum of the dc self-bias and the plasma potential (approximately V in this system). Therefore even though the ion energy is low at high source powers, the ion flux is sufficient to create significant degradation in the rectifier characteristics. The forward voltage drop for a Schottky rectifiers is related to the barrier height (B) and RON through the relation [56]: JRJVFONBFFn T AenkT)ln(2** where k is Boltzmanns constant, T the absolute temperature, e the electronic charge, JF the forward current density and A** is Richardsons constant for 4H-SiC. Therefore VF can be degraded by increases in ideality factor and on-state resistance along with a reduction in forward current due to introduction of trap states. The changes in rectifier performance were also dependent on the applied rf chuck power, or equivalently on ion energy. The changes in VB, VF, RON and n were 20% for VB and VF, 30% for RON and 40% for n over the range of these parameters that we investigated. Figure 3-10 shows the percentage changes in VB (top) and RON (bottom) as a function of process pressure at fixed source power (300 W) and rf power (100 W). Note that the decreases in both parameters are largest at the lowest pressure. This is consistent with the fact that at low pressures the incident ions have a lower probability of collisions with gas molecules or they traverse the sheath region and therefore they impact with their full energy. A major advantage of the plasma ICP source is its ability to operate at higher

PAGE 67

54 pressures ( 40 mTorr) relative to the more normal 10mTorr range of conventional cylindrical coil sources. The ideality factor of the rectifiers was still more degraded at higher pressures, while VF showed very small changes with pressure (Figure 3-11). 3.3.4 Summary and Conclusions In conclusion 4H-SiC Schottky rectifiers show only modest changes in the electrical characteristics after high-energy proton irradiation at fluences corresponding to more than 10 years in low-earth orbit. These devices appear promising for both aerospace and terrestrial applications where irradiation hardness is a prerequisite. The main degradation mechanism appears to be creation of recombination centers and traps that cause an increase in reverse leakage current, ideality factor and on-state resistance. In another study, PECVD SiO2 layers were deposited on Ni/4H-SiC Schottky rectifiers as a function of various plasma parameters. Both reverse breakdown voltage and on-state resistance increase with increasing plasma power and SiH4 content through creation of deep trap states and degradation of the SiC surface by reaction with atomic hydrogen, respectively. The changes in forward turn-on voltage are less dramatic in each case. Plasma conditions that utilize low-to-moderate powers, high process pressure and low SiH4 percentages produce minimal changes in the rectifier performance. 4H-SiC power rectifiers were also exposed to Ar discharges in a plasma coil ICP reactor as a function of various plasma parameters. The reverse breakdown voltage decreases with increasing ion flux, which is controlled by the source power, whereas VF, RON and n all increase under the same conditions. These results are consistent with creation of surface states that promote reverse breakdown at lower applied voltages. The same trends were observed with increasing of chuck power, which controls the incident ion energy. The increases in VB and RON are minimized at high operating process

PAGE 68

55 pressure, as expected since these conditions reduce the effectiveness of the ion bombardment experienced during the plasma exposure. Damage during actual etch or deposition process would be lower than produced during the Ar plasma exposure, but clearly process conditions that utilize low ion energies and fluxes are desirable at the end of an etch cycle or the beginning of a deposition cycle in order to minimize ion-induced damage.

PAGE 69

56 ICP (Inductively Coupled Plasma) Etching System Reactor 13.56 MHZ Frame Structure Faraday Shield Load Lock He Cooling System Plasma Gas Inlet Quartz Sample Powered Electrode Figure 3-7 Planar coil ICP reactor.

PAGE 70

57 100200300400500600700-25-20-15-10-5 SiC Schottky RIE Power: 100 WPressure: 10 mTorr ICP Power (W)% change in VB0100200300400500 dc self-bias (-V)100200300400500600700-2000200400600800100012001400 RIE Power: 100 WPressure: 10 mTorrSiC Schottky % change in RonICP Power (W) Figure 3-8 Percentage change in VB (top) and RON (bottom) of 4H-SiC rectifiers as a function of ICP source power.

PAGE 71

58 100200300400500600700050100150200250 RIE Power: 100 WPressure: 10 mTorrSiC Schottky % change in VFICP Power (W)1002003004005006007001020304050607080 RIE Power: 100 WPressure: 10 mTorrSiC Schottky % change in ideality factorICP Power (W) Figure 3-9 Percentage change in VF (top) and n (bottom) of 4H-SiC rectifiers as a function of ICP source power.

PAGE 72

59 10152025303540-20-15-10-505 Pressure (mTorr)% change in VB050100150200250 SiC SchottkyRIE Power: 100 WICP Power: 300 Wdc self-bias (-V)1015202530354014161820222426283032 RIE Power: 100 WICP Pwer: 300 WSiC Schottky % change in RonPressure (mTorr) Figure 3-10 Percentage change in VB (top) and RON (bottom) of 4H-SiC rectifiers as a function of process pressure.

PAGE 73

60 10152025303540171819202122 RIE Power: 100 WICP Pwer: 300 WSiC Schottky % change in VFPressure (mTorr)10152025303540202530354045505560 RIE Power: 100 WICP Pwer: 300 WSiC Schottky % change in ideality factorPressure (mTorr) Figure 3-11 Percentage change in VF (top) and n (bottom) of 4H-SiC rectifiers as a function of process pressure.

PAGE 74

CHAPTER 4 JUNCTION TERMINATION EXTENSION GEOMETRY OF SIC RECTIFIERS It is possible to make etching of SiC (0001) facet only by plasma methods because of its high chemical stability [128]. After such etching the surface of the semiconductor is strongly damaged and is not electrically neutral. It is a reason for arising breakdown at considerable lower voltages than the values implied by the impurity concentration in the base. In most of the cases the breakdown characterized by surface leakage current (in the order of 10-6-8 A) and was irreversible. For decreasing probability of surface like breakdown it is necessary to make surface electrical field strength less than the strength in bulk. This can be achieved by profiling of mesa structure during etching [129] or by diffusion o acceptor impurities into the surface region of the n-base [130]. Also electrical activity of the surface can be decreased by special treatment like oxidation. Another problem encountered for high reverse voltage is when the breakdown occurs through the air along the surface of the mesa-structure. It has been shown that the breakdown voltage increases when the measurements are done by placing the samples in a dielectric liquid with a high value of electric breakdown field strength like fluorientTM FC-77 [131]. For high temperature and high power devices if is necessary to cover mesa-structure by the thermostable dielectric with critical electrical field strength more than that of air. 61

PAGE 75

62 4.1 Influence of Edge Termination Geometry on Performance of 4H-SiC P-i-N Rectifiers 4.1.1 Introduction 4H-SiC has a larger bandgap (3.26 eV) compared to 6H-SiC (2.9 eV) and is preferred for device applications because of its higher and more isotropic electron mobility [70]. There is particular interest in the development of 4H-SiC power rectifiers for high power electronics applications. The on-state resistance of SiC rectifiers can be less than 1% that of Si rectifiers with the same breakdown voltage [106-108]. There are two basic classes of rectifiers, namely Schottky and p-n junction. The former have lower on-state voltages and higher switching speeds, while the latter have higher breakdown voltage and lower reverse leakage current. Advanced designs such as junction barrier Schottky (JBS) [109], dual metal trench [90] and merged p-i-n Schottky rectifiers (MPS) [110] attempt to achieve the advantages of both types of rectifiers [84]. An empirically derived relationship of 4H-SiC shows that the critical electric field for breakdown, EC, is related to the doping level in the drift layer, ND, through [112]: )10log(25.01105.2166DCNE or in terms of breakdown voltage, VB [104]: N V DB75.0151075.1 Excellent high current (230 A) and high voltage (5.5 kV) performance have been reported for 4H-SiC p-i-n rectifiers. The performance of manufacturable devices in the 300-1000V range is of particular interest, especially the development of robust designs and fabrication sequences. In this paper we report on the influence of edge termination design on the performance of 4H-SiC p-i-n rectifiers with a simple epi layer structure.

PAGE 76

63 4.1.2 Experimental Methods Epi layers of 12 m of n-type (n~2x1015 cm-3) and 1 m of p-type (p~1017 cm-3) 4H-SiC were grown on n+ 4H-SiC substrates. A schematic of the completed rectifiers is shown in Figure 4-1. A full-area back metallization of Ni annealed at 970 0C for 3 min provided ohmic contact to the substrate. Mesas were formed by dry etching and thermal SiO2 was used as the part of the dielectric overlap edge termination. The e-beam evaporated Ni Schottky rectifiers had areas between 0.005-0.04 mm2. Passivation layers of SiO2 (3000 thick) were deposited by plasma enhanced chemical vapor deposition. The adjustable parameters were the extent of metal overlap onto the mesa defining the active device area and the length of this mesa. The mesa edge termination has been demonstrated to provide excellent results on SiC p-n junctions provided the mesa surface is properly passivated [90,94]. 4.1.3 Results and Discussion Figure 4-2 shows the forward I-V characteristics from one of the p-i-n rectifiers. The I-V curve can be fitted to the standard relationship for p-n diodes, namely [107]: )exp()exp(21kTneVJkTneVJAIJDORO where A is the diode area, JRO and JDO are the saturation current densities for recombination and diffusion current mechanisms, respectively, e is the electronic charge, k is Boltzmanns constant, T the absolute temperature, V the applied bias and n1 and n2 are the ideality factors for recombination and diffusion currents, respectively. A fit to the data yields n1=1.97 and n2=1.10. Therefore, at lower current densities the transport is dominated by recombination, whereas at higher forward voltages, the current is

PAGE 77

64 dominated by the diffusion contribution. The forward turn-on voltage, VF, can be represented as [84]: VVMiFnnnekT)ln(2 where nand n+ are the electron concentrations at the two end regions (p+/n and n+/n), ni the intrinsic electron concentration and VM the voltage drop across the drift region. The measured VF is ~4V, or roughly 0.74 V higher than the theoretical minimum. The best on-state resistance was 15 mcm2, with an on/off current ratio of 1.5x105 at 3/-450 V. Figure 4-3 shows reverse I-V characteristics as a function of metal overlap distance for fixed mesa extension distance of 60 m. The reverse breakdown voltage shows a trend of decreasing as the metal overlap distance increases. The mechanism is not clear at the moment, but could be due to several factors, including breakdown in the thermal oxide or extension of the depletion region to include defects under the mesa. The figure-of-merit VB2/RON was a maximum of 13.5 MWcm-2 for these rectifiers. Figure 4-4 shows the variation of VB with metal overlap distance for rectifiers with different active areas. There is no significant effect of area over the range we employed, but we would expect to observe a decrease in VB in very large rectifiers because of the increased probability of having a defect in the active region [127]. Figure 4-5 shows reverse I-V characteristics from rectifiers with fixed area and extent of metal overlap, as a function of mesa extension length. The VB values extracted from this data are shown in Figure 4-6. Within experimental error and the uniformity of VB on a given wafer (~ %), there is no significant change in breakdown voltage as a function of mesa extension length.

PAGE 78

65 4.2 Effect of Contact Geometry on 4H-SiC Schottky Rectifiers with Junction Termination Extension 4.2.1 Introduction There is great current interest in SiC power Schottky rectifiers with reverse breakdown voltages in the 500-1000V range for applications in traction motor control, sensors and next generation communications and radar systems [125-132]. These devices have significant advantages over Si diodes for these purposes, including lower switching losses, higher operating temperatures and faster switching speeds [93, 96]. In addition the high thermal conductivity of SiC (approximately three times that of Si) makes it ideal for very high current applications. There is still a need to optimize the edge termination and contact geometry for SiC rectifiers in order to find the most robust design for manufacturing. In this study we report on the effect of Schottky metal overlap and contact shape on the reverse breakdown voltage of 4H-SiC rectifiers employing junction termination extension (JTE). The JTE is used to reduce the electric field at the edges of the rectifiers. It increases the breakdown voltage of the diode by reducing the electric field density within the SiC near the edges of the rectifier. The p-doping of the JTE region counteracts the bending of the depletion region around the edges of the anode. This effect spreads out the electric field at the corners and edges of the depletion region. 4.2.2 Experimental Methods The starting samples consisted of 1 m p-type 4H-SiC (p~1017 cm-3) on 12 m of n-type 4H-SiC (n~2x1015 cm-3) on a n+ 4H-SiC substrate. A full area back contact of e-beam deposited Ni annealed at 970 0C was used for contact to the substrate. Thermally

PAGE 79

66 Figure 4-1 P-i-N rectifier.

PAGE 80

67 012345670.02.0x10-34.0x10-36.0x10-38.0x10-31.0x10-21.2x10-21.4x10-2 SiC P-i-N diode Current (A)Bias Voltage (V) Figure 4-2 Forward I-V characteristics of p-i-n rectifier.

PAGE 81

68 -700-600-500-400-300-200-100010-1110-1010-910-810-710-610-510-4 metal overlap 0 m metal overlap 2 m metal overlap 4 m metal overlap 6 mSiC P-i-N diodeSiO2: 3000 p-SiC:1 mDiode area: 0.01 mm2mesa length:60 m Current (A)Bias Voltage (V) Figure 4-3 Reverse I-V characteristics from p-i-n rectifiers as a function of metal overlap distance onto mesa.

PAGE 82

69 01234560200400600 area: 0.04 mm2 area: 0.01 mm2 area: 0.005 mm2 Breakdown Voltage (V)Metal Overlap (m) Figure 4-4 Variation of VB with metal overlap distance for p-i-n rectifiers.

PAGE 83

70 -500-450-400-350-300-250-200-150-100-50010-1110-1010-910-810-710-610-510-4 mesa length: 60 m mesa length: 40 m mesa length: 20 m mesa length: 0 mSiC P-i-N diodeSiO2: 3000 p-SiC:1 mDiode Area: 0.04 mm2metal overlap: 4 m Current (A)Bias Voltage (V) Figure 4-5 Reverse I-V characteristics from p-i-n rectifiers as a function of mesa length.

PAGE 84

71 0102030405060100150200250300350400450 SiC P-i-N diodeSiO2: 3000 p-SiC: 1 mArea: 0.04 mm2 Metal Overlap: 4 m Breakdown Voltage (V)Mesa Length (m) Figure 4-6 Variation of VB with mesa length for p-i-n rectifiers with fixed metal overlap distance of 4 m.

PAGE 85

72 grown SiO2 was on part of the dielectric overlap edge termination. Schottky contact holes were etched into the SiO2 and through the p-SiC to expose the underlying n-SiC using dry etching. Schottky contacts of e-beam evaporated Ni (area 0.04 0.64 mm2) were patterned by lift-off. Three different contact shapes were employed, namely oval, circular and square. Finally, SiO2 layer deposited by plasma enhanced chemical vapor deposition (PECVD) was used for passivation. A schematic of the completed device structure is shown in Figure 4-7. 4.2.3 Results and Discussion The forward turn-on voltage VF, defined as the forward bias at which the current density is 100 A-cm2, was < 2V (Figure 4-8). The on-state resistance, RON, was 4.2mcm2, which is close to the theoretical minimum for this structure of 4H-SiC. The VF is ~25% higher than the theoretical minimum for Ni Schottky diodes on 4H-SiC and may reflect the presence of some residual dry etch damage or surface contamination. Figure 4-9 shows the reverse current-voltage (I-V) characteristics from rectifiers with 0.04 mm2 area, junction termination extension of 40 m and Ni metal overlap of 4 m in extent, as a function of contact shape. Note that the oval diodes have the highest leakage at low bias (<250 V) and the lowest leakage at higher voltages. The reverse breakdown increases in the sequence square
PAGE 86

73 approximately a factor of two lower than expected from the depletion layer doping and thickness employed. Figure 4-10 shows reverse I-V characteristics from rectifiers of different contact area and fixed JTE (40 m) and extent of metal overlap (2 m). The reverse breakdown voltage decreases with increasing contact area. This is most likely due to the increased probability of having crystal defects in the active region of the rectifier as the area is increased. The shape of the I-V curves is similar to those reported earlier for Ni/4H-SiC rectifiers [90], where it was suggested that both thermionic field emission and field emission mechanisms are present. There was little effect of metal overlap extent on the reverse I-V characteristics, as shown in Figure 4-11 for oval-shaped rectifiers with small area (0.04 mm2) and a JTE of 20 m. This is expected, since the field distribution in the depletion region is controlled by the p-n junctions formed by the JTE regions on the drift region and not by the metal overlying the p-regions. This gives considerable latitude in the patterning step for the metal deposition. Similar results are shown in Figure 4-12 for rectifiers with a JTE length of 40 m. A summary of VB data as a function of metal overlap distance for different JTE length is shown in Figure 4-13. While the number of samples is too small to draw definitive conclusions, VB is maximized at shorter JTE lengths. We believe that breakdown may actually be initiated in the p-layer under some conditions because of its low doping level and therefore longer JTE regions have a higher probability of having defects that promote breakdown. The diodes show an excellent on/off current ratio (1.5/-500V) of 4x105 at 25 0C.

PAGE 87

74 4.3 Role of Device Area, Mesa Length and Metal Overlap Distance on Breakdown Voltage of 4H-SiC P-i-N Rectifiers 4.3.1 Introduction SiC p-i-n rectifiers have several advantages over their simpler Schottky counterparts, most importantly, a higher reverse breakdown voltage (VB) and lower reverse leakage current [104-110]. SiC p-i-n rectifiers are expected to offer significant performance advantages over Si rectifiers for high voltage applications, rf base station infrastructure, advanced avionics, combustion control electronics in automobiles and solid-state power conditioning [133-135]. While very high VB and forward conduction currents have been reported (8.6 kV [87]), the need is to design robust device structures and processes for the 300-1200 V range. In particular, to realize a manufacturable SiC rectifier process, it is necessary to understand the effect of various edge termination geometries on the breakdown voltage characteristics so that VB most closely approaches the expected parallel-plate value for the epi layer doping and thicknesses employed. Moreover, it is necessary to carry out these investigations on 4H-SiC, which has larger bandgap and higher and more isotropic mobility than the 6H-polytype [70]. In this study we report the results of an investigation into the role of device area, mesa length and metal overlap distance on the breakdown voltage of 4H-SiC p-i-n rectifiers fabricated on thick (~50 m) epi-layers under non-punch through drift region conditions. 4.3.2 Experimental Methods The starting substrates were n+ (1019 cm-3) 4H-SiC. Approximately 50 m of n-type (~2 x 1015 cm-3) and 1 m of p-type (p ~ 1017 cm-3) 4H-SiC was grown on these substrates by vapor phase epitaxy, followed by ~500 of thermal SiO2. A full area-back

PAGE 88

75 Figure 4-7 Schottky rectifier.

PAGE 89

76 0.00.51.01.52.0020406080 SiC Schottky Diode(p-type edge termination)SiO2: 3000 Diode Area: 0.64 mm2 Current Density (A/cm2)Bias Voltage (V) Figure 4-8 Forward I-V characteristics of 0.64 mm2 rectifier.

PAGE 90

77 -800-700-600-500-400-300-200-100010-1110-1010-910-810-710-610-510-4 oval diode circular diode square diodeSiC Schottky Diodes area:0.04 mm2JTE:40 mmetal overlap: 4 m Current (A)Bias Voltage (V) Figure 4-9 Reverse I-V characteristics of 0.04 mm2 rectifiers with different contact shape.

PAGE 91

78 -700-600-500-400-300-200-100010-1110-1010-910-810-710-610-5 area: 0.04 mm2 area: 0.16 mm2 area: 0.64 mm2SiC Schottky Diodes ovalJTE:40 mmetal overlap: 2 m Current (A)Bias Voltage (V) Figure 4-10 Reverse I-V characteristics of rectifiers with oval shaped contacts of different area.

PAGE 92

79 -800-700-600-500-400-300-200-100010-1110-1010-910-810-710-610-5 metal overlap: 6 m metal overlap: 4 m metal overlap: 2 mSiC Schottky Diodes (Oval)area:0.04 mm2JTE:20 m Current (A)Bias Voltage (V) Figure 4-11 Reverse I-V characteristics of rectifiers with oval shaped contacts (0.04 mm2), as a function of extent of metal overlap.

PAGE 93

80 -900-800-700-600-500-400-300-200-100010-1110-1010-910-810-710-610-510-4 SiC Schottky Diodes (Oval)area:0.04 mm2JTE:40 m metal overlap: 2 m metal overlap: 4 m metal overlap: 6 m Current (A)Bias Voltage (V) Figure 4-12 Reverse I-V characteristics of rectifiers with oval-shaped contacts (0.04 mm2), as a function of extent of metal overlap.

PAGE 94

81 23456620640660680700720740760780800820 SiC Schottky Diodes area:0.04 mm2 JTE: 60 m JTE: 40 m JTE: 20 mBreakdown Voltage (V)Metal Overlap (m) Figure 4-13 Variation of VB with extent of metal overlap for different JTE lengths.

PAGE 95

82 contact of e-beam deposited Ni annealed at 970 0C for 3 min (contact resistivity of 1.2 x 10-6 cm2), was used as the ohmic contact to the substrate. SiO2 layers ~ 3000 were deposited by plasma enhanced chemical vapor deposition (PECVD) and were employed as part of dielectric overlap edge termination. Mesas were formed by dry etching. Holes were also opened in the SiO2 stack by a combination of wet and dry etching and Schottky contacts of e-beam evaporated (1000 thick) Ni patterned by lift-off. The contact area was varied from 0.01-0.36 mm2, while the mesa length was varied from 0-60 m and the extent of metal overlap from 0-20 m. All current-voltage (I-V) measurements were carried out at 25 0C in air. 4.3.3 Results and Discussion Figure 4-14 shows reverse I-V characteristics as a function of rectifier active area for a fixed mesa length of 20 m and a metal overlap distance of 4 m. The reverse breakdown voltage shows a trend of decreasing as the area increases. This is most likely a result of the greater probability for having defects such as micropipes within the active region of the rectifier as the area increases. Therefore, our devices still have breakdown characteristics dominated by defect density rather than bulk breakdown mechanisms. For the smallest rectifiers the best on-state resistance (RON) obtained was ~ 20 mcm2, leading to a maximum power figure-of-merit VB2/RON of 84.5 MW.cm-2 This is roughly a factor of 6 higher than we reported previously for 4H-SiC p-i-n rectifiers fabricated on thinner (~ 12 m) n-layers [136]. The dependence of VB on diode area is shown in Figure 4-15 for fixed mesa length of 20 m and metal overlap distance of 4 m. The fall-off in VB with diode area is most pronounced at small areas (0.01-0.04 mm2) and most likely defines the region where

PAGE 96

83 there is a probability of having either zero or one killer-defect within the device. Compared to the theoretical best performance for 4H-SiC rectifiers of this structure, there is still considerable improvement possible in both VB and RON [84]. The reverse I-V characteristics as a function of metal overlap distance for fixed mesa extension length of 0 m and fixed diode area of 0.04 mm2 are shown in Figure 4-16. VB goes through a maximum with increasing metal overlap distance, which is a somewhat different trend than reported previously where it continuously decreased as the extension distance increased [136]. However, a difference in that case was the mesa length was fixed at 60 m [136]. In the present case, the trend may result from onset of breakdown in the oxide or extension of the depletion region to include defects that initiate breakdown. Figure 4-17 shows the variation of VB with metal overlap distance for the same diodes whose I-V characteristics are shown in Figure 4-16. It is clear that this parameter has a strong influence on the breakdown voltage, with VB ranging from a low of ~ -800 V to a maximum of ~ -1200 V depending on metal overlap distance. The reverse I-V characteristics from rectifiers with fixed area and extent of metal overlap are shown in Figure 4-18 as a function of mesa extension length. Over a wide range of voltage, the reverse current density was found to vary as the square root of the reverse voltage, which is characteristic of space-charge generation currents. The relation between generation current Jo and effective generation lifetime ( G) is given by: 5.05.0)()2(VneNeJGiBG

PAGE 97

84 where e is the electronic charge, the permittivity of 4H-SiC, NB the drift layer doping concentration, ni the extrinsic carrier concentration at the measurement temperature and V is the applied reverse bias [87]. Given a typical value of G =100 ns [87], this relation yields a calculated generation current much lower than the measured value, suggesting that other mechanisms such as surface leakage is playing a role. The larger area devices should lower reverse current densities, indicating that leakage around the contact periphery is the main contributor, as reported previously [87]. Figure 4-19 shows the variation of VB with mesa length, extracted from the data of Figure 4-18. There is no clear dependence of reverse breakdown on this parameter under our conditions. Finally, Figure 4-20 shows a typical forward I-V characteristic from a small area rectifier. The I-V characteristic is well represented by the relation: )()(21kTneVDkTneVReIeII where IR and ID are the saturation currents for recombination and diffusion current mechanisms, respectively, e is the electronic charge, k is the Boltzmanns constant, V the applied voltage, T the absolute temperature and n1 and n2 are the ideality factors for recombination and diffusion currents, respectively. At low currents, n ~ 2, indicating that the dominant transport mechanism is recombination. At higher currents, n ~ 1-1.5, indicating that the dominant mechanism is diffusion. A typical on/off current ratio was ~ 104 at +3.5V/-1000V. 4.3.4 Summary and Conclusions In conclusion mesa edge termination of p-n 4H-SiC rectifiers passivated with SiO2 is simple alternative to use of implanted junction termination extension (JTE) methods. The mesa extension length did not have a significant impact on the reverse breakdown

PAGE 98

85 voltage of our devices, whereas increasing metal overlap distance tended to decrease VB. On/off current ratios >105 were achieved, with VF values ~25% higher than the theoretical minimum. In another study, 4H-SiC rectifiers were fabricated with epi JTE regions and dielectric overlap for edge termination. Contact areas from 0.04-0.64 mm2 were examined, with the reverse breakdown voltage being inversely dependent on contact area. Oval and circular diodes showed higher VB than square contacts of the same area. The metal overlap extent did not have a strong influence on VB for JTE lengths up to 40 m. 4H-SiC p-i-n rectifiers with VB ~ -1000 V were also demonstrated. The role of device area, mesa length and metal overlap distance on VB were examined. The breakdown voltage was inversely dependent on device area but was not a function of mesa length. On/off current ratios of ~ 104 were achieved at 3.5/-1000V. VB was a maximum at ~ 5 m metal overlap distance.

PAGE 99

86 -1000-800-600-400-20010-1110-910-710-5 area: 0.36 mm2 area: 0.16 mm2 area: 0.04 mm2 area: 0.01 mm2SiC p-i-n diodesMesa length: 20 mMetal overlap: 4 m Current (A)Bias Voltage (V) Figure 4-14 Reverse I-V characteristics of p-i-n rectifiers as a function of active area.

PAGE 100

87 0.00.10.20.30.40200400600800100012001400 SiC p-i-n diodesMesa length: 20 mMetal Overlap: 4 m VB (-V)Diode Area (mm2) Figure 4-15 VB as a function of diode area for p-i-n rectifiers with mesa length 20 m and metal overlap 4 m.

PAGE 101

88 -1400-1200-1000-800-600-400-20010-1110-910-710-5 Metal overlap: 0m Metal overlap: 2m Metal overlap: 5m Metal overlap: 10m Metal overlap: 20mSiC p-i-n diodesDiode area: 0.04 mm2Mesa length: 0 m Current (A)Bias Voltage (V) Figure 4-16 Reverse I-V characteristics of p-i-n rectifiers as a function of metal overlap distance.

PAGE 102

89 051015200200400600800100012001400 SiC p-i-n diodesDiode area: 0.04 mm2 Mesa length: 0 m VB (-V)Metal Overlap (m) Figure 4-17 VB as a function of metal overlap length for p-i-n rectifiers with area 0.04 mm2 and zero mesa length.

PAGE 103

90 -1200-1000-800-600-400-20010-1110-910-710-5 Mesa length: 0m Mesa length: 20m Mesa length: 40m Mesa length: 60mSiC p-i-n diodeDiode area: 0.04 mm2Metal overlap: 0 m Current (A)Bias Voltage (V) Figure 4-18 Reverse I-V characteristics of p-i-n rectifiers as a function of mesa length.

PAGE 104

91 01020304050600200400600800100012001400 SiC p-i-n diodesDiode area: 0.04 mm2Metal overlap: 0 m VB (-V)Mesa Length (m) Figure 4-19 VB as a function of mesa length for p-i-n rectifiers with area 0.04 mm2 and zero metal overlap length.

PAGE 105

92 0123450.06.0x10-41.2x10-31.8x10-32.4x10-33.0x10-33.6x10-3 SiC p-i-n diodeDiode area: 0.04 mm2Mesa length: 40 mMetal overlap: 5 m Current (A)Bias Voltage (V) Figure 4-20 Forward I-V characteristics.

PAGE 106

LIST OF REFERENCES 1. J.W. Palmour, J.A. Edmond, H.S. Kong and C.H. Carter, Silicon Carbide Related Material, 137 499 (1992). 2. K. Shenai, R.S. Scott and B.J. Baliga, IEEE Electron Dev. Lett., 36 1811 (1989). 3. B.J. Baliga, IEEE Electron Dev. Lett., 10 455 (1989). 4. T.P. Chow and R. Tyagi, IEEE Trans. Electron Devices, 41 1481 (1994). 5. A. Bhalla and T.P. Chow, Proc. 6th International Symp. Power Semiconductor Devices and ICs, 287 53 (1994). 6. T.P. Chow, N. Ramangul and M. Ghezzo, Mat. Res. Soc. Symp. Proc., 483 89 (1998). 7. R. Held, N. Kaminski and E. Niemann, Mat. Sci. Forum, 264 1057 (1998). 8. G.J. Campisi, Proc. IEEE, 1238 56 (2000). 9. J.A. Cooper and A. Agarwal, Proc. IEEE, 90 6 (2002). 10. Y. Sugawara, D. Takayama, K. Asano, R. Singh. J. Palmour and T. Hayashi, Proc. 13th International. Symp. Power Semiconductor Devices and ICs, 27 45 (2001). 11. J. Tan, J.A. Cooper, Jr., and M.R. Melloch, IEEE Electron Dev. Lett. 19 487 (1998). 12. S.-H. Ryu, A.K. Agarwal, R. Singh, and J.W. Palmour, IEEE Electron Dev. Lett. 22 124 (2001). 13. H. Lendenmann, N. Johansson, D. Mou, M. Frischholz, B. strand, P. Isberg, and C. Ovren, Mater. Sci. Forum, 338 1423 (2000). 14. Technical Literature (Cree Research Inc., Durham, NC, 2001). 15. A. Ellison, T. Kimoto, I.G. Ivanov, Q. Wahab, A. Henry, O. Kordina, J. Zhang, C.G. Hemmingsson, C.Y. Gu, M.R. Leys and E. Janzen, Mater. Sci. Forum, 103 264 (2000). 93

PAGE 107

94 16. N.V. Dyakonov, P.A. Ivanov, V.A. Kozlov, M.E. Levinshtein, J.W. Palmour, S.L. Rumyantsev and R. Singh, IEEE Trans. Electron Devices, 46 2188 (1999). 17. R. Held, M. Fullmann and E. Niemann, Mater. Sci. Forum, 338 1407 (2000). 18. R. Rupp, M. Treu, A. Mauder, E. Griebl, W. Werner, W. Bartsch and D. Stephani, Mater. Sci. Forum, 338 1167 (2000). 19. J.B. Cascady and R.W. Johnson, Solid State Electron., 39 1409 (1996). 20. M.V. Rao, P. Griffiths, O.W. Holland, G. Kelner, J.A. Freitas, D.S. Simons, P.H. Chi and M. Ghezzo, J. Appl. Phys., 77 2479 (1995). 21. H. Wirth, D. Panknin, W. Skorupa and E. Niemann, Appl. Phys. Lett., 74 979 (1999). 22. T. Kimoto, N. Inoue and H. Matsunami, Phys. Status Solidi A, 162 263 (1997). 23. T. Ohshima, A. Uedono, H. Itoh, M. Yoshikawa, K. Kojima, S. Okada, I. Nashiyama, K. Abe, S. Tanigawa, T. Frank and G. Pensl, Mater. Sci. Forum, 338 857 (2000). 24. M.A. Capano, J.A. Cooper, M.R. Melloch, A. Saxler and W.C. Mitchel, Mater. Sci. Forum, 338 703 (2000). 25. P.H. Chi, E.M. Handy and M.V. Rao, proceedings of SIMS, Brussels, 234 79 (1999). 26. S.E. Saddow, J. Williams, T Issac-Smith, M. Capano, J.A. Cooper, M.S. Mazzola. A.J. Hsieh and J.B. Cascady, Mater. Sci. Forum, 338 901 (2000). 27. M. Lazar, L. Ottaviani, M.L. Loccateli, D. Planson, B. Canut and J.P. Chante, Mater. Sci. Forum, 338 921 (2000). 28. Y. Tanaka, N. Kobayashi, H. Okumura, R. Suzuki, T. Ohdaira, M. Hasegawa, M. Ogura, S. Yoshida and H. Tanoue, Mater. Sci. Forum, 338 909 (2000). 29. J.B. Tucker, E.M. Handy, M.V. Rao, O.W. Holland, N. Papanicolaou, and K.A. Jones, Mater. Sci. Forum, 338 925 (2000). 30. K. Chatty, V. Khemka, T.P. Chow and R.J. Gutmann, Mater. Sci. Forum, 338 1331 (2000). 31. J.A. Cooper, Phys. Status Solidi A, 162 305 (1997).

PAGE 108

95 32. V.V. Afanasev, Microelectron. Eng., 48 241 (1999). 33. L.A. Lipkin and J.W. Palmour, J. Electron. Mater., 25 909 (1996). 34. L.A. Lipkin and J.W. Palmour, IEEE Trans. Electron. Devices, 46 525 (1999). 35. M. Bassler, G. Pensl and V. Afanasev, Diam. Relat. Mater., 6 1472 (1997). 36. T. Ouisse, Phys. Status Solidi A, 162 339 (1997). 37. M.K. Das and J.A. Cooper, Mater. Sci. Forum, 338 1069 (2000). 38. R. Schrner, P. Friedrichs and D. Peters, IEEE Trans. Electron Devices, 46 533 (1999). 39. C.J. Anthony, A.J. Jones and M.J. Uren, Mater. Sci. Eng. B: Solid-State Mater. Adv. Technol., 61 460 (1999). 40. M.M. Maranowski and J.A. Cooper, IEEE Trans. Electron Devices, 46 520 (1999). 41. M. Bassler, V. Afanasev, G. Pensl and A.M. Schulz, Microelectron. Eng., 48 257 (1999). 42. A.O. Konstantinov, Q.Wahab, N. Nordell and A.U. Lindefelt, J. Electron. Mater., 27 335 (1998). 43. J. Crofton, L.M. Porter and J.R. Williams, Phys. Status Solidi B, 202 581 (1997). 44. J. Crofton, L. Beyer, J.R. Williams, E.D. Luckowski, S.E. Mahoney and J.M. Delucca, Solid-State Electron, 41 1725 (1997). 45. S. Liu, G. Potts and J. Schofield, Mater. Sci. Forum, 338 1021 (2000). 46. T. Jang, G. Rutsch, B. Odekirk and L.M. Porter, Mater. Sci. Forum, 338 1001 (2000). 47. L. Kassamakova, R. Kakanakov, I. Kassamakov, N. Nordell, S. Savage, E.B. Svedberg and L.D. Madsen, Mater. Sci. Forum, 338 1009 (2000). 48. D.J. Morrison, A.J. Pidduck, V. Moore, P.J. Wilding, K.P. Hilton, M.J. Uren and C.M. Johnson, Mater. Sci. Forum, 338 1199 (2000). 49. T. Kimoto, Q. Wahab, A. Ellison, U. Forsberg, M. Tuominen, R. Yakimova, A. Henry and E. Janzen, Mater. Sci. Forum, 264 921 (1998).

PAGE 109

96 50. Q. Wahab, T. Kimoto, A. Ellison, C Hallin, M. Tuominen, R. Yakimova, A. Henry, J.P. Bergman and E. Janzen, Appl. Phys. Lett., 72 445 (1998). 51. H. Mitlehner, W. Bartsch, M. Bruckmann, K.O. Dohnke and U. Weinert, Proceeding of ISPSD97, 165 59 (1997). 52. S. Ortolland, M.L. Locatelli,, D. Planson, J.P.Chante and A. Senes, Mater. Sci. Forum, 264 1045 (1998). 53. D. Alok and B.J. Baliga, IEEE Trans. Electron Devices, 44 1013 (1997). 54. K.J. Schoen, J.M. Woodall, J.A. Cooper and M.R. Melloch, IEEE Trans. Electron Devices, 45 1595 (1998). 55. A.P. Knights, M.A. Lourenco, K.P. Homewood, D.J. Morrison, N.G. Wright, S. Ortolland, C.M. Johnson, A.G. ONeill and P.G. Coleman, J. Appl. Phys., 87 3973 (2000). 56. D.T. Morisette, J.A. Cooper, Jr., M.R. Melloch, G.M. Dolny, P.M. Shenoy, M. Zafrani and J. Gladish, IEEE Trans. Electron Devices, 48 349 (2001). 57. J.B. Casady, in Processing of Wide Bandgap Semiconductors, ed. S.J.Pearton (William Andrews Publishing, Norwich, NY, 2000). 58. G. Pensl and W.J. Choyke, Physics B, 185 264 (1993). 59. T.P. Chow and M. Ghezzo, Mat. Res. Soc. Symp. Proc., 423 9 (1996). 60. S. Sriram, G. Augustine, A.A. Burk, R.C. Glass, H.M. Hobgood, P.A. Orphanos, L.B. Rowland, T.J. Smith, C.D. Brandt, M.C. Driver and R.H. Hopkins, IEEE Electron Dev. Lett., 17 369 (1996). 61. J.B. Casady, J.D. Cressler, W.C. Dillard, R.W. Johnson, A.K. Agarwal and R.R. Siergiej, Solid-State Electron., 39 777 (1996). 62. J.N. Shenoy, J.A. Cooper, Jr. and M.R. Melloch, IEEE Electron Dev. Lett., 18 93 (1997). 63. P.G. Neudeck and C. Fazi, IEEE Electron Dev. Lett., 18 96 (1997). 64. K. Xie, J.H. Zhao, J.R. Flemish, T. Burke, W.R. Buchwald, G. Lorenzo and H. Singh, IEEE Electron Dev. Lett., 17 325 (1996). 65. O. Kordina, J.P. Bergman, A. Henry, E. Janzen, S. Savage, J. Andre, L.P. Ramberg, V. Lindefelt, W. Hermansson and K. Bergman, Appl. Phys. Lett., 61 1561 (1995).

PAGE 110

97 66. Y. Luo, L. Fursin and J.H. Zhao, Electron. Lett., 36 1496 (2000). 67. S.H. Ryu, A.K. Agarwal, R. Singh and J.W. Palmour, IEEE Electron Dev. Lett., 23 124 (2001). 68. A.Q. Huang and B. Zhang, Solid State Electron., 44 325 (2000). 69. M.A. Capano, R. Santhakumar, R. Venugopal, M.R. Melloch and J.A. Cooper, Jr., J. Electron. Mater., 29 210 (2000). 70. J.B. Casady, A.K. Agarwal, S. Seshadri, R.R. Siergiej, L.B. Rowland, M.F. MacMillan, D.C. Sheridan, P.A. Sanger and C.D. Brandt, Solid-State Electron., 42 2165 (1998). 71. J.B. Casady and R.W. Johnson, Solid-State Electron., 39 1409 (1996). 72. A.K. Agarwal, S. Seshadri, M. MacMillan, S.S. Mani, J. Casady, P. Sanger and P. Shah, Solid-State Electron., 44 303 (2000). 73. Y.M. Sung, J.B. Casady, J.B. Dufrene and A.K. Agarwal, Solid-State Electron., 46 605 (2002). 74. S.E. Mohney, B.A. Hull, J.Y. Lin and J. Crofton, Solid-State Electron., 46 689 (2002). 75. M. Tarplee, V. Madangarli and T.S. Sudarshan, Solid State Electron, 46 753 (2002). 76. P.A. Ivanov, M.E. Levinshtein, S.L. Rumyantsev, S. Ryu, A.K. Agarwal and J.W. Palmour, Solid-State Electron, 46 567 (2002). 77. S.P. Murray and K.P. Roenker, Solid State Electron., 45 1847 (2001). 78. Q. Zhang and T.S. Sudarshan, Solid-State Electron., 39 1235 (2000). 79. Q. Zhang and T.S. Sudarshan, Solid-State Electron., 45 1805 (2001). 80. Y. Gow, Y. Tang, M. Hoshi and T.P. Chow, Solid-State Electron., 44 1875 (2000). 81. D.C. Sheridan, G. Niu, J.N. Merrett, J.D. Cressler, C. Ellis and C.C. Tin, Solid-State Electron., 44 1367 (2000). 82. T.P. Chow, V. Khemka, J. Fedison, N. Ramungul, K.Matocha, Y. Tang and R.J. Gutmann, Solid-State Electron., 44 277 (2000).

PAGE 111

98 83. A.Q. Huang and B. Zhang, Solid-State Electron., 44 325 (2000). 84. V. Khemka, R. Patel, T.P. Chow and R.J. Gutmann, Solid-State Electron., 43 1945 (1999). 85. J.A. Cooper, Jr., M.R. Melloch, R. Singh, A. Agarwal and J. Palmour, IEEE Trans. Electron Dev., 49 545 (2002). 86. M.C. Tarplee, V.P. Madangarli, Q. Zhang and T.S. Sudarshan, IEEE Trans. Electron Dev., 27 435 (2001). 87. R. Singh, J.A. Cooper, Jr., M.R. Melloch, T.P. Chow and J.W. Palmour, IEEE Trans. Electron Dev., 49 665 (2002). 88. D.T. Morisette, J.A. Cooper, Jr., M.R. Melloch, G.A. Dolny, P.M. Shenoy, M. Zafrani and J. Gladish, IEEE Trans. Electron Dev., 46 2188 (2001). 89. N. Dyakonova, P.A Ivanov, V.A. Kozlov, M.E. Levinshtein, J.W. Palmour, S.L. Rumyantsev and R. Singh, IEEE Trans. Electron Dev., 46 2188 (1999). 90. K.J. Schoen, J.M. Woodall, J.A. Cooper, and M.R. Melloch, IEEE Trans. Electron Dev., 45 1595 (1998). 91. P.G. Neudeck, J. Electron. Mater., 24 283 (1995). 92. C.E. Weitzel, J.W. Palmour, C.H. Carter, Jr., K. Moore, K.K. Nordquist, S. Allen, C. Thero and M. Bhatnagar, IEEE Trans. Electron Dev., 43 1732 (1996). 93. V. Saxena, J.N. Su and A.J. Steckl, IEEE Trans. Electron Dev., 46 456 (1999). 94. D. Alok and B.J. Baliga, IEEE Electron Dev. Lett., 44 1013 (1997). 95. J. Crofton, P.G. McMullin, J.R. Williams and M.J. Bozack, J. Appl. Phys., 77 1317 (1995). 96. B.J. Baliga, Power Semiconductor Devices (PWS, Boston, MA, 1996). 97. T.S. Marinova, A. Kakanakova-Georgieva, V. Krastev, R. Kakanakov, M. Neshev, L. Kassamakova, O. Noblanc, C. Ornodo, S. Cassette, C. Brylinski, B. Pecz, G. Radnoczi, Gy. Vincze, Mater. Sci. Engineering B, 46 223 (1997). 98. T.C. Chou, A. Joshi and J. Wadsworth, J. Vac. Sci Technol., A9 1525 (1991). 99. J. Crofton, P.G. MacMullin, J.R. Williams and M.J. Bozack, J. Appl. Phys., 77 1317 (1995).

PAGE 112

99 100. M.W. Cole, P.C. Joshi, C.W. Hubbard, M.C. Wood, M.H. Ervin, B. Geil and F. Ren, J. Appl. Phys, 88 2652 (2000). 101. J. Crofton, L.M. Porter and J.R. Williams, Phys. Status Solidi B, 202 581 (1997). 102. A. Bachli, M.A. Nicolet, L. Baud, C. Jaussaud and R. Madar, Mater. Sci. Eng., B, 56 11 (1998). 103. C.S. Pai, C.M. Hanson and S.S. Lau, J. Appl. Phys., 57 618 (1985). 104. M.C. Tarplee, V.P. Madangarli, Q. Zhang and T.S. Sudarshan, IEEE Trans. Electron Dev., 48 2659 (2001). 105. P. Alexandrov, J.H. Zhao, W. Wright, M. Pan and M. Weiner, Electron. Lett., 37 1261 (2001). 106. M. Trivedi and K. Shenai, J. Appl. Phys., 85 6889 (1999). 107. D.Peters, R. Schrner, K-H. Hlzlein and P. Friedrichsm Appl. Phys.Lett., 71 2996 (1997). 108. C.I. Harris and A.O. Konstantinov, Physica Scripta, T79 27 (1999). 109. B.J. Baliga, IEEE Electron Dev. Lett., 5 194 (1984). 110. M. Mehrotra and B.J. Baliga, Solid-State Electron., 38 703 (1995). 111. B.J. Baliga and H.R. Chang, IEEE IEDM Tech. Dig., 658 661 (1987). 112. A.O. Konstantinov, N. Nordell, Q. Wahab and U. Lindefelt, Appl. Phys. Lett., 73 1850 (1998). 113. Q. Zhang and T.S. Sudarshan, J. Electron. Mater., 30 1466 (2001). 114. D.J. Morrison, N.G. Wright, A.B. Horsfall, C.M. Johnson, A.G. O'Neill, A.P. Knights, K.P. Hilton and M.J. Uren, Solid-State Electron., 44 1879 (2000). 115. A. Itoh, T. Kimoto, and H. Matsunami, IEEE Elctron Dev. Lett., 17 139 (1996). 116. M.G. Rastegaeva, A.N. Andreev, A.A. Petrov, A.I. Babanin, M.A. Yagovkina and I.P. Nikitina, Mat. Sci. Eng., B46 254 (1997). 117. V. Saxena and A.J. Steckl, in SiC Devices, ed. R.K. Willardson and E.R. Weber (Academic Press, San Diego, CA, 1998).

PAGE 113

100 118. A. Itoh, T. Kimoto, and H. Matsunami, IEEE Elctron. Dev. Lett., 16 280 (1995). 119. K.V. Vassilevski, A.B. Horsfall, C.M. Johnson, N.G. Wright and A.G. O'Neill, IEEE Trans. Electron. Dev., 49 947 (2002). 120. E. Wendler, A Heft and W. Wesch, Nucl. Instr. Meth. in Physics Research, B141 105 (1998). 121. W.J. Choyke, Int. Conf. Rad. Effects in Semiconductors (IUPAP, Dubrovnik), 58 69 (1997). 122. M.V. Rao, Solid State Electron., (in press). 123. S. Mitra, M.V. Rao, K. Jones, N Papanicoulov and S. Wilson, Solid State Electron., (in press). 124. T. Chen, Z. Luo, J.D. Cressler, T.F. Isaacs-Smith, J.R. Williams, G. Chung and S.D. Clark, Solid State Elctron., (in press). 125. J.W. Corbett and J.C. Bourgoin, in Point Defects in Solids, ed. J.H. Crawford and F.F. Slifkin (Plenum, NY, 1975). 126. M.C. Tarplee, V.P. Madangarli, Q. Zhang and T.S. Sudarshan, IEEE Trans. Electron Dev., 49 665 (2002). 127. M.G. Spencer, J. Palmour and C. Carter, IEEE Trans. Electron. Dev., 49 940 (2002). 128. A.L. Syrkin, I.V. Popov and V.E. Chelnokov, Sov. Tech. Phys. Lett., 12 99 (1986). 129. F. Lanous, D. Planson, M.L. Locatelli, J.P. Chante, Appl. Phys. Lett.,23 385 (1998). 130. A.A. Lebedev, A.N. Andreev, A.A. Maltsev, M.G. Rastegaeva, N.S. Savkina and V.E. Chelnokov, Semiconductors, 29 850 (1995). 131. P.G. Neudeck, D.J. Larkin, J.A. Powell, L.G. Matus and C.S. Salupo, Appl. Phys. Lett., 64 1384 (1994). 132. W. Wright, J. Carter, P. Alexandrov, M. Fan, M. Weiver and J.H. Zhao, Electron. Lett., 37 787 (2001). 133. B.J. Baliga, Proc. IEEE, 89 822 (2001). 134. K. Satoh and M. Yamamoto, Proc. IEEE, 89 813 (2001).

PAGE 114

101 135. A. Lidow, D. Kinzer, G. Sheridan and D. Tam, Proc. IEEE, 89 803 (2001). 136. S. Nigam, Jihyun Kim, B. Luo, F. Ren, G.Y. Chung, K. Shenai, P.G. Neudeck, S.J. Pearton and J.R. Williams, Solid-State Electron. (in press).

PAGE 115

BIOGRAPHICAL SKETCH Saurav Nigam was born April 13th, 1978, in the small village of Rewa in central India. His family moved to Bhopal in 1981 and he started his schooling at Campion School, Bhopal, where he spent his fourteen years before graduating from high school in 1996. In 1997, he began his college at the Indian Institute of Technology (IIT), Roorkee, one of the premier institutions in Asia. He graduated from IIT in May 2001.During the undergraduate studies he conducted research in the area of biochemical engineering in collaboration with Dr. Subhash Chand, professor, IIT-Delhi. He spent the summer of 1999 at the Department of Biotechnology and Biochemical Engineering at IIT-Delhi conducting experiments on Lipase Catalyzed Esterification Reactions in Supercritical CO2. It was his strong interest in the courses on reaction engineering, material science and its application and electrical engineering that led to his ever-increasing interest in microelectronics. He enrolled at the University of Florida in the Department of Chemical Engineering in August 2001. In the spring of 2002 he joined Prof. Fan Rens research group to work on state-of-art compound semiconductor devices. He has worked on silicon carbide based devices focusing on Schottky and P-i-N rectifiers for high power electronics. He has also published more than ten papers in reputable refereed journals and symposium proceedings during his graduate and undergraduate research. 102


Permanent Link: http://ufdc.ufl.edu/UFE0000758/00001

Material Information

Title: Silicon Carbide Schottky and P-I-N Rectifiers
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0000758:00001

Permanent Link: http://ufdc.ufl.edu/UFE0000758/00001

Material Information

Title: Silicon Carbide Schottky and P-I-N Rectifiers
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0000758:00001


This item has the following downloads:


Full Text











SILICON CARBIDE SCHOTTKY AND P-I-N RECTIFIERS


By

SAURAV NIGAM


















A THESIS PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF SCIENCE

UNIVERSITY OF FLORIDA


2003




























Copyright 2003

by

Saurav Nigam






















To my parents, my fiancee, my brother and my friends for their endless love and support.















ACKNOWLEDGMENTS

The past year has been one of the most crucial years of my life because for the first

time in my professional career, I was introduced to an exciting realm of microelectronics.

I would like to thank my advisor, Prof Fan Ren, for giving me this golden opportunity to

join his research group. His commitment to provide world-class facilities in his

laboratory, engage students in the cutting-edge research and above all to provide

excellent guidance in all the spheres of life, both professional and personal, is truly

exceptional. Needless to mention, his extreme hard work and determination will always

act as a source of inspiration to me and instill in me the confidence to face every

adversity in life. He has always been able to find some time out of his extremely busy

schedule to help me to steer my efforts in the right direction. I will always look forward

to him as my true mentor.

I am also indebted to the other members of my advisory committee: Steve Pearton

and Tim Anderson. Their significant contributions to this work are greatly appreciated.

During the collaboration with Prof Steve Pearton's and Prof. Cammy Abernathy's

research groups, I had an opportunity to develop interpersonal skills, emphasizing team

efforts and it was here that I came across some of the most remarkable people, like Brent

Gila and Mark, that I have ever known.

I would like to thank Brent for always lending me a helping hand whenever I

needed him and for his perspicacious views especially during the Wednesday meetings. I









will also always cherish nice little discussions with Mark, who has provided me his

considerable time and energy to help me to set up the dicing saw.

I would also like to thank Ben Luo not just for providing the guidance in research

but also for providing his invaluable advice and guidance. I will always cherish those

moments when he would come up with his nice little humorous comments that would

take away all the stress after the day's work. I would always count on him as a truly

sincere friend to seek for the right guidance. I would also like to thank Rishabh for his

support and care. He has been patient in providing me his valuable time to gain hands-on

experience with various equipment together with some delicious food and recipes. I have

also learned a lot from Jihyun, who has collaborated most closely with my research. I

would also like to thank Yoshi for providing me help and support and always showing a

great deal of concern. His friendship is one of the greatest gifts to me. Kelly, Kwang and

Kyu-pil have always been kind enough to lend their helping hand with the research.

I would like to thank Dennis Vince and Jim Hinnat of the Chemical Engineering

Department's machine shop for all their help. I would also like to thank the staff in the

Chemical Engineering Department for all their administrative help. They are Peggy-Jo

Daugherty, Sonja Pealer, Santiago, Nancy Krell, Janice Harris and Debbie Sadoval. Their

assistance allowed me to focus on research.

I would like to thank my parents, my brother and my fiancee for their endless love,

care and support. They have provided me the strength and inspiration to continue my

studies at the University of Florida. This acknowledgement cannot be complete without

thanking my roommates Gopal, Nori, Sudeep and Archit, who have been very

understanding and caring throughout my stay at Gainesville.
















TABLE OF CONTENTS
page

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

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

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

ABSTRACT ........ .............. ............. .. ...... .......... .......... xii

CHAPTER

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

1.1 B ackgrou n d .................... ...... ......... ......................... ............................ . 1
1.2 Current Status of SiC Power Technology............. ......... ............... 4
1.3 SiC Semiconductor Devices for Power Applications .................... ............ 7
1.3.1 M material A availability and Quality ........................................ .............. 8
1.3.2 SiC Processing Technology ........................... ....... ........................ 9
1.3.2 .1 Ion im plantation ................................................... ... ... .... .. ..... 10
1.3 .2 .2 O xides on SiC ............................................................ ..... .... 11
1.3.2.3 Contacts ............. ................................................. ........ 12
1.3.2.4 Edge termination and passivation................................... ............... 13

2 SILICON CARBIDE SCHOTTKY RECTIFIERS ............................................. 14

2.1 Experimental Characteristics of 4H-SiC Schottky Rectifiers............................ 14
2.1.1 Introduction ........... ... ........ ............... 14
2.1.2 Experimental Methods ............. .... ............ .............. 15
2 .1.3 R results and D iscu ssion ........................................................ .... .. .............. 17
2.2 N i Contacts to N -Type 4H -SiC .................... ... ........................... ...26
2.3 Stability of SiC Schottky Rectifiers to Rapid Thermal Annealing .................. 27
2.3.1. Introduction .................. ....... ....... ........ 27
2.3.2 Experim ental M ethods ....................................................... ......... ..... 29
2 .3.3 R results and D iscu ssion ........................................................ .... .. .............. 29
2.3.4 Sum m ary and C onclusions.................................................. ... ................. 32

3 IRRADIATION AND PASSIVATION EFFECTS ON 4H-SiC RECTIFERS ...........38

3.1 High Energy Proton Irradiation Effects on SiC Schottky Rectifiers .................... 38
3.1.1 Introduction .............. ...... ..... ..... ...... ............ ........... 38









3.1.2 E xperim ental M ethods ......................................................................... ... 39
3.1.3 Results and Discussion............................. .............. ....... 39
3.2 Influence of PECVD of Si02 Passivation Layers on 4H-SiC Schottky Rectifiers41
3.2.1 Introduction ................................. .......................... .... ....... 41
3.2.2 Experim ental M ethods ........................................................ ......... ..... 45
3.2.3 Results and Discussion........................... ... ... .... ................. 45
3.3 Effect on 4H-SiC Schottky Rectifier of Ar Discharges Generated in a Planar
Inductively Coupled Plasma Source ............. ................................ .............. 47
3.3.1 Introduction ........... ... ........ ............... 47
3.3.2 Experimental Methods ............. .... ............ .............. 51
3.3.3 R results and D iscussion............................................ .......................... 51
3.3.4 Sum m ary and C onclusions.................................................. ... ................. 54

4 JUNCTION TERMINATION EXTENSION GEOMETRY OF SiC RECTIFIERS..61

4.1 Influence of Edge Termination Geometry on Performance of 4H-SiC P-i-N
Rectifiers ............... ..... ... ...... .............. ........ .... .......... 62
4.1.1 Introduction ........... ... ........ ............... 62
4.1.2 Experimental Methods ............ ..... ........ .................. 63
4.1.3 Results and D iscussion.................................. ......... ... ........ ............. 63
4.2 Effect of Contact Geometry on 4H-SiC Schottky Rectifiers with Junction
Term nation Extension ................................................ .......... .............. 65
4.2.1 Introduction ................................. ....................................... 65
4.2.2 Experim ental M ethods ....................................................... ......... ..... 65
4.2.3 R results and D iscussion............................ .... ..... ........ .............. 72
4.3 Role of Device Area, Mesa Length and Metal Overlap Distance on Breakdown
Voltage of 4H-SiC P-i-N Rectifiers .............................................. ........... 74
4.3.1 Introduction ............... ........ ............... 74
4.3.2 Experimental Methods .............. ......................................... 74
4 .3.3 R results and D iscu ssion ........................................................ .... .. .............. 82
4.3.4 Sum m ary and C onclusions.................................................. ... ................. 84

LIST OF REFEREN CES ..................................................................... ............... 93

BIO GR A PH ICA L SK ETCH .................................... ........... ......................................102
















LIST OF TABLES


Table page

1-1 Physical properties of important semiconductors for high voltage power devices. .5

1-2 Normalized unipolar figures of merit of important semiconductors for high
voltage pow er devices .................. ......................... .. ......... ... ........ ....
















LIST OF FIGURES


Figure page

1-1 Comparison of the ideal breakdown voltage of Si and SiC devices for different
doping levels. ................................................ .......................... 2

1-2 Comparison of the blocking layer thickness as a function of the ideal breakdown
voltage for SiC and Si. ........................ ................... ................. 3

2-1 Cross-section of SiC Schottky rectifiers (top) and photograph of completed 2
inch diam eter w afer (bottom ).......................................... ........................... 16

2-2 Reverse I-V characteristics at 25 OC from Schottky rectifiers with different
contact diam eters. .................... ................ .................................18

2-3 Reverse I-V characteristic from 154 |tm diameter rectifier (top) and map of
reverse breakdown voltage from a quarter of a 2 inch diameter wafer (bottom). .19

2-4 Comparison of breakdown voltage obtained from simulations and the fabricated
d ev ic e s................................. ......................................................... ............... 2 1

2-5 Calculated reverse breakdown voltage as a function of epilayer doping ..............23

2-6 Forward I-V characteristics in linear (top) and log (bottom) scales.....................24

2-7 Specific on-resistance as a function of breakdown voltage...............................25

2-8 Contact resistivity of Ni ohmic contact on n+ 4H-SiC as a function of annealing
temperature (top), annealing time (center) and annealing ambient (bottom) ......28

2-9 Forward I-V characteristics as a function of anneal temperature for 60 sec
an n e a ls .......................................................................... 3 3

2-10 Change in RON as a function of anneal temperature for 60 sec anneals. ................34

2-11 Forward I-V characteristics (top) and change in VF (center) and RON (bottom)
as a function of anneal time at 900 C. ...................................... ............... 35

2-12 Optical micrographs of Ni contacts before (top) and after annealing at 1000C
for 60 sec (center) or 120 sec (bottom). ....................................... ............... 36









2-13 Change in VF (top) and RON (bottom) as a function of measurement temperature
after 900C, 60 sec anneals. ..... ....................................................................... 37

3-1 Reverse I-V characteristics from 4H-SiC Schottky rectifiers before and after
proton irradiation at a dose of 5 x 109 cm-2 (top) and percentage increase in
reverse leakage at -250V (bottom)......................................................................42

3-2 Forward I-V characteristics from 4H-SiC Schottky rectifiers before and after
proton irradiation at a dose of 5 x 109 cm-2 (top) and percentage decrease in
forward current at 2V (bottom) ................................................................. ....... 43

3-3 Numerical change in values of n (top), RON (center) and VF (bottom) as a
function of proton fluence................................................................................. 44

3-4 Percentage change in VB as a function of plasma power (top), process pressure
(center) and N20 content (bottom) during SiO2 deposition..............................48

3-5 Percentage change in VF as a function of plasma power (top) and process
pressure (bottom) during SiO2 deposition................................... ............... 49

3-6 Percentage change in RON as a function of plasma power (top), N20 content
(center) and process pressure (bottom) during SiO2 deposition ..........................50

3-7 Planar coil ICP reactor ....................................................................... 56

3-8 Percentage change in VB (top) and RON (bottom) of 4H-SiC rectifiers as a
function of ICP source pow er. ................. ........ ............ ................ ....... 57

3-9 Percentage change in VF (top) and n (bottom) of 4H-SiC rectifiers as a function
of ICP source pow er .......... .. ........................ ........ ..... ...... .. ............ 58

3-10 Percentage change in VB (top) and RoN (bottom) of 4H-SiC rectifiers as a
function of process pressure.......................................................................... ... 59

3-11 Percentage change in VF (top) and n (bottom) of 4H-SiC rectifiers as a function
of process pressure.................... ................ ...................... .......... 60

4 -1 P -i-N rectifi er ....................... .. ...............................................................6 6

4-2 Forward I-V characteristics of p-i-n rectifier. ............. ............ ....................... 67

4-3 Reverse I-V characteristics from p-i-n rectifiers as a function of metal overlap
distance onto m esa. .................... .................... .............................68

4-4 Variation of VB with metal overlap distance for p-i-n rectifiers. .........................69

4-5 Reverse I-V characteristics from p-i-n rectifiers as a function of mesa length......70









4-6 Variation of VB with mesa length for p-i-n rectifiers with fixed metal overlap
d istan c e o f 4 tm ....................................................................................7 1

4-7 Schottky rectifier. ........................................... ................. .... ....... 75

4-8 Forward I-V characteristics of 0.64 mm2 rectifier................................................76

4-9 Reverse I-V characteristics of 0.04 mm2 rectifiers with different contact shape. ..77

4-10 Reverse I-V characteristics of rectifiers with oval shaped contacts of different
area ................................................................................. 7 8

4-11 Reverse I-V characteristics of rectifiers with oval shaped contacts (0.04 mm2),
as a function of extent of metal overlap........... .......... ..... ................79

4-12 Reverse I-V characteristics of rectifiers with oval-shaped contacts (0.04 mm2),
as a function of extent of metal overlap ................... ........ ................. .. ....... 80

4-13 Variation of VB with extent of metal overlap for different JTE lengths ................81

4-14 Reverse I-V characteristics of p-i-n rectifiers as a function of active area.............86

4-15 VB as a function of diode area for p-i-n rectifiers with mesa length 20 |tm and
m etal overlap 4 m .................... ........................................ ............... 87

4-16 Reverse I-V characteristics of p-i-n rectifiers as a function of metal overlap
distance. ..................................................................88

4-17 VB as a function of metal overlap length for p-i-n rectifiers with area 0.04 mm2
and zero m esa length............ ..... ... ............................... ....................89

4-18 Reverse I-V characteristics of p-i-n rectifiers as a function of mesa length...........90

4-19 VB as a function of mesa length for p-i-n rectifiers with area 0.04 mm2 and zero
m etal overlap length ............. .. .... .............. ...... ..... .... .............. 91

4-20 Forward I-V characteristics. ............................................................................. 92















Abstract of Thesis Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Master of Science

SILICON CARBIDE SCHOTTKY AND P-I-N RECTIFIERS

By

Saurav Nigam

May 2003


Chair: Fan Ren
Major Department: Chemical Engineering

There has been a significant research interest in silicon carbide (SiC) over the past

few years as a base material system for high frequency and high power semiconductor

devices. SiC Schottky diodes have shown an excellent performance for the realization of

power diodes for fast switching applications with nearly negligible power loss. In spite of

extremely high switching speeds, Schottky diodes suffer from high leakage current. P-i-N

diodes offer low leakage currents but show reverse recovery charge during switching and

have a large junction forward voltage drop due to the wide bandgap of 4H-SiC. However,

both the diodes have shown a considerable improvement in maximum breakdown voltage

values compared to that of planar junctions, with the employment of proper edge

termination methods.

The dc performance of Si02 dielectric overlap-terminated 4H-SiC Schottky

rectifiers was simulated for different drift layer doping levels and dielectric thickness.

The devices were then fabricated with different Schottky contact diameters and the dc









results compared to the simulations. There was a strong dependence of reverse

breakdown voltage (VB) on contact diameter (4) ranging from -750 V for 100 [tm ) to -

440 V for 1000 |tm. The reverse leakage current scaled with diode diameter, indicating

that the surface contributions are dominant. The specific on-state resistance was -0.36

mQcm2, close to the theoretical minimum, with a forward turn-on voltage of 1.82 V at

100 A.cm-2. The contact resistivity of back-side Ni contacts annealed at 970 OC was -1.2

x 10-6 Qcm2.

The effect of deposition conditions of plasma enhanced chemical vapor deposition

(PECVD) SiO2 layers and inductively coupled Ar plasmas on the electrical properties of

4H-SiC Schottky rectifiers was also studied. For the former, the changes in reverse

breakdown voltage (VB), forward turn-on voltage (VF) and on-state resistance (RoN) were

< 20% over a broad range of plasma conditions and show that 4H-SiC is relatively

resistant to changes induced by the ion bombardment and hydrogen flux present during

PECVD of dielectrics to surface passivation. In the latter, the VB increased by increase in

both incident ion energy and ion flux.

4H-SiC P-i-N rectifiers with SiO2 passivated mesa edge termination showed

forward current characteristics dominated by recombination at low bias (n-1.97) and

diffusion at high voltages (n-1.1). The forward turn-on voltage was -4V, with a specific

on-state resistance of 15 mQcm2, on/off current ratio of 1.5x105 at 3V/-450V and figure-

of-merit, VB2/RoN, of 13.5 MWcm-2. The mesa extension distance did not have a strong

impact on reverse breakdown voltage.














CHAPTER 1
INTRODUCTION

1.1 Background

Silicon carbide (SiC) has shown outstanding potential for high power, high

temperature, high frequency and high voltage devices used in advanced communications

and radar system sensor and control systems, utility power switching and traction motor

control. The material has excellent physical and electrical properties such as extremely

high thermal conductivity (up to 4.9 W/cmK), high breakdown electric field (2-4 x 106

V/cm), a wide bandgap (3.26 eV for 4H-polytype) and high electron saturation velocity

(2 x 107 cm/s) as shown in Table 1-1 [1]. These characteristics result in significantly

higher current densities, operating temperatures and breakdown voltage than comparable

Si devices and in lower switching and on-state losses.

A comparison of ideal breakdown voltages versus blocking layer doping

concentration (Figure 1-1) suggests that the more highly doped blocking layer (more than

10 times higher) provides lower resistance for SiC because more majority carriers are

present than for comparably rated Si devices. A comparison of voltage blocking layer

thickness for a given breakdown voltage (Figure 1-2) shows a thinner blocking layer (n-

-5 x 1015 cm-3) of SiC devices (1/10th that of Si devices) also contribute to the lowering of

specific on-resistance by a factor of 10. The combination of 1/10th the blocking layer

thickness with 10 times the doping concentration can yield a SiC device with a factor of

100 advantage in resistance compared to that of Si devices.











104



103



102



101


1014


1015


1016


Doping of the blocking layer (cm3)


Figure 1-1 Comparison of the ideal breakdown voltage of Si and SiC devices for different
doping levels.


Si











f
S103
ISSi


S 10






*OM
10 10 104

Ideal breakdown voltage (V)


Figure 1-2 Comparison of the blocking layer thickness as a function of the ideal
breakdown voltage for SiC and Si.
Ida rakonvltg V

Fiue omprsno h lciglyrtikesa ucino h da









Several figures of merit, specifically to quantify the intrinsic performance

potentials of unipolar and bipolar devices, have been proposed [2-5]. The most important

material parameter is the avalanche electric field, followed by thermal conductivity and

carrier mobility. The unipolar figures of merit are shown in Table 1-2 [1] where SiC and

A1N unipolar devices have been projected to have intrinsic device improvement of 80 to

30000 times over conventional Si [4]. Also, bipolar devices, such as pin junction rectifier

and GTO thyristors, are expected to have lower total power loss when the switching

frequencies exceed 1 and 30 kHz respectively [5]. Here, it is worth pointing out that BN

is the best among the Group III-nitrides for power devices because of its high thermal

conductivity and indirect bandgap. Furthermore, due to the higher carrier mobility and

more isotropic nature of its properties, 4H-SiC has become the polytype of choice [6].

1.2 Current Status of SiC Power Technology

The overall goal for power electronic circuits is to reduce power losses, volume,

weight, and, costs of the system [7]. These factors become significant for applications in

future military and commercial ships especially electric propulsion. The first SiC based

device was a light emitting diode fabricated in 1907 by H. J. Round. However, a more

intensive silicon carbide semiconductor material and device development has been

underway for nearly 35 years in United States, the former Soviet Union, Europe and

Asia. Only in the past 15 years with the input of substantial funding from the US DOD

community (BMDO, DARPA, and ONR) have we seen the development of SiC wafers

industry in the US able to produce single crystal substrates with lower defect density

capable of sustaining device development programs [8]. Today, there are three suppliers

of SiC wafers in US (CREE, Sterling and Litton Airtron), one in Europe and one in Japan

with additional companies evaluating the market potential [8]. Within the last five years,















Table 1-1 Physical properties of important semiconductors for high voltage power
devices.
Material E, n, ,. /., Ec V0, Direct
eV cm"3 cm:/V-s 106V/cm 107cm/s W/cm-K Indirect
Si 1.1 1.5x010' 11.8 1350 0.3 1.0 1.5 1
Ge 0.66 2.4Ax013 16.0 3900 0.1 0.5 0.6 1
GaAs 1.4 1.8xl06 12.8 8500 0.4 2.0 0.5 D
GaP 23 7.7x10 11.1 350 1.3 1.4 0.8 I
InN 1.86 -10' 9.6 3000 1.0 25 D
GaN 3.39 1.9x10 10 9.0 900 3.3 2.5 1.3 D
3C-SiC 22 6.9 9.6 900 1.2 2.0 4.5 I
4H-SiC 3.26 8.2xl0-9 10 72 2.0 2.0 4.5 I
650t
6H-SiC 3.0 2.3x100 9.7 3 2.4 2.0 4.5 I
50C
Diamond 5.45 1.6x10-27 5.5 1900 5.6 2.7 20 I
BN 6.0 15x10 3' 7.1 5 10 1.0* 13 I
AIN 6.1 -10-3 8.7 1100 11.7 1.8 2.5 D
Note: a mobility along a-axis. c mobility along c axis, *- estimate.


Table 1-2 Normalized unipolar figures of merit of important semiconductors for high
voltage power devices.
Material X JM MJM KM Qn Q2 BM (Q3) BHFM
(Ev./7)2 *JM 4(v3/s,)"' X(A u 1E, s,,E? LE
Si I 1 1 1 1 1 1 1
Ge 0.4 0.03 0.012 0.20 0.06 0.02 0.2 03
GaAs 0.33 7.1 2.4 0.45 5.2 6.9 15.6 10.8
GaP 0.53 37 20 0.7 10 40 16 5
InN 58 46 19
GaN 0.87 760 655 1.6 560 6220 650 77.8
3C-SiC 3 65 195 1.6 100 400 33.4 103
4H-SiC 3 180 540 4.61 390 2580 130 22.9
6H-SiC 3 260 780 4.68 330 2670 110 16.9
Diamond 13.3 2540 33800 32.1 54860 1024000 4110 470
BN 8.7 1100 9700 11 715 23800 83 4
AIN 1.7 5120 8700 21 52890 2059000 31700 1100
Note: a mobility along a-axis, c mobility along c axis









commercial devices and applications have emerged. Specifically, SiC is used for blue and

green LEDs, Northrop-Grumman makes Static Induction Transistors for internal

applications, and CREE has announced their recent SiC Zero RecoveryTM Rectifiers.

Due to its exceptional switching speed, a high operational temperature and a high

breakdown electric field, SiC is having a profound impact on the design, topology and

circuits used in the power electronics especially where space weight, power density and

thermal management issues dominate as in mobile platforms; on aircraft, on vehicles and

especially on ships.

There have been many funded programs carried out at universities, industry and

government laboratories to commercialize SiC power switching devices. These programs

are designed to address device processing and fabrication technology, device physics and

design, packaging and applications.

The most recent results in Schottky and PiN diodes demonstrated exceptional

switching speeds with little or no ringing. When SiC diodes were packaged with Si

switches to form a simple hybrid inverter circuit, nearly all the stored energy was

eliminated due to the fast turn-off/turn-on SiC diodes [8].

At this time, US has a large but distribute program in SiC power device technology,

sustained with DOD programs at universities and larger companies such as GE and

Northrop Grumman. Traditional semiconductor solid-state switch manufacturers are now

actively pursuing this technology. On the other hand, the more integrated European

power industry, ABB and Siemens are actively pursuing research in SiC technology at all

levels in an internally coordinated manner. In Asia, Japan has an active research program

in SiC power technology that is now becoming focused under a new initiative [8].









1.3 SiC Semiconductor Devices for Power Applications

Silicon carbide offers significant advantages for power-switching devices because

the critical field for avalanche breakdown is about ten times higher than in silicon. SiC

power devices have made remarkable progress in the past five years, demonstrating

currents in excess of 100 A and blocking voltages in excess of 19000 V [9]. Over the past

decade a number of new SiC-based power switching devices have been demonstrated

such as GTOs, MOSFETs, pn diodes and schottky diodes with results approaching or

exceeding the theoretical limits of Si. However, there is significant difference between

demonstrating prototype devices and establishing an economically viable commercial

product. Silicon has proven to be an extremely difficult material to displace, largely

because of economic factors, and it remains to be seen whether SiC can accomplish what

other semiconductor materials have failed to do.

There are significant new challenges emerging, both in material science and device

design as the SiC technology is developed. One of the unique features of SiC is its high

thermal stability compared to silicon. This has both advantages as well as disadvantages-

an advantage because of improved reliability and higher temperature operability, and a

disadvantage because many of the typical fabrication processes either do not occur at all

in SiC or require extremely high temperatures. The most challenging aspect of the

thermal stability is the fact that due to the phase equilibrium in the Si-C system, SiC does

not melt, but instead gradually sublimes at temperatures above 2000 K. This makes it

impossible to form large single-crystal ingots by pulling a seed crystal from a melt, as in

Czochralski process that produces 200-300 mm diameter silicon ingots. Instead a

modified sublimation process that is presently limited to 100 mm diameter ingots forms

SiC crystals. The process is expensive and difficult and SiC wafer costs are astronomical









compared to silicon. However, the advantages of SiC power devices are so great that the

higher cost of the starting material can still be offset in certain cases, allowing SiC to

compete directly with silicon in specialized device markets.

The evolution of SiC power-switching technology can be divided into three phases:

prototype device demonstration, device scale-up, and commercial production. Many

types of SiC power devices are well along in the prototype demonstration phase and

several have demonstrated performance measures far superior to silicon. These include

rectifiers with blocking voltages above 19 kV [10], power metal-oxide-semiconductor

field-effect transistors (MOSFETs) [11], and bipolar junction transistors (BJTs) [12] with

performance figures 100 times higher than in silicon, and gate turn-off (GTO) thyristors

blocking over 3 kV and switching over 60 kW. Several of these devices are in the scale-

up phase, with terminal currents in excess of 100 A already demonstrated in single-device

packaged rectifiers [13].

1.3.1 Material Availability and Quality

SiC has a large number of different crystallographic forms or poly-types. Of these,

6H-SiC has the most developed growth technology on account of its relatively large

volume usage as a substrate for GaN blue LEDs. However, 4H-SiC is the preferred

material for many power applications owing to its nearly ten times higher on-axis

mobility compared to 6H-SiC. At present commercial 4H-SiC wafers are available in

sizes up to 50 mm diameter while 6H-SiC wafers are available in sizes up to 75mm

diameter, with 100 mm at the research stage. Substrates are available in both low

resistivity (n- and p-type) and semi-insulating forms [14]. However, the moderate to high

resistivity substrates desirable for high voltage devices is not available. Devices are

therefore, fabricated on homoepitaxial layers, which are routinely grown to thicknesses of









over 100 |tm with doping densities as low as 1014 cm-3 using hot wall chemical vapor

deposition (CVD) [15]. High-level carrier lifetimes for this material are typically of the

order of several hundred nanoseconds [16] making it suitable for the fabrication of

devices with voltage blocking capabilities approaching 10 kV.

Historically, the major difficulty with SiC has been the presence of micropipes in

the substrates and epilayers. It is generally conceded that micropipes are the major

obstacles in the production of high-performance SiC devices. Micropipes are defects

unique to the growth of SiC and they are physical holes that penetrate through the entire

crystal. They are replicated into the device epitaxial layers and become "killer" defects if

they intersect the active region of SiC devices. However, the reports on material growth

with micropipe densities as low as 0.1/cm2 at research level (reduced from over 1000/cm2

in just a few years) [14], indications are that SiC is now adequate to fabricate devices

several millimeters square with reasonable yield [17, 18]. Indications are that a

dislocation density of less than 103 cm-2 is desirable for fabrication of power devices [18]

(current values are typically around 104 cm-2). A further difficulty concerns the

uniformity of epilayer doping and thickness across the wafer (typical results show 4 %

standard deviation in doping) and the uniformity of doping between runs (typically 40%)

[14].

1.3.2 SiC Processing Technology

SiC has a relatively mature processing technology. All the basic process steps, such

as doping (by implantation and during epitaxy), etching (plasma techniques), oxide

growth, Schottky and ohmic contacts have been successfully demonstrated [19]. There









are, however, several key processing issues that currently limit the performance of

fabricated devices.

1.3.2.1 Ion implantation

Ion implantation is employed widely for local p-type and n-type doping of SiC. The

most frequently used implanted ions are aluminum (Al) and boron (B) for p-type,

nitrogen (N) and phosphorous (P) for n-type. Key issues in SiC implantation technology

include the reduction of lattice damage occurring during implantation, successful

electrical activation of the dopants and preservation of good surface morphology.

Performing the implantation at elevated temperature is a common way to reduce the

lattice damage and reduce the requirements for subsequent annealing [20]. Typical

implantation temperatures are: 400 C for Al [21], 500 C for N [22], 800 C for P [23].

These temperatures are sufficiently high to obtain good electrical conductivity after

annealing but not so high as to cause surface dissociation and/or large vacancy clusters

[24].

The annealing condition must be precisely determined, both for recording the

crystal and diffusing the dopants into substantial sites. In addition, evaporation of both

silicon and dopants must be avoided if the surface stoichiometry is to be preserved. This

is particularly important in cases where the surface layer forms an active part of the

device, for example in MOSFET and MESFET channel regions. There are two main

approaches to preserve the surface: (i) material encapsulation with A1N, graphite [25]; (ii)

increased Si partial pressure in the furnace reactor, either with silane gas [26] or with SiC

coated pieces [27]. Several authors have given results for optimum electrical activation

[28-30]. In all cases, it appears that good electrical activation is only possible if the

annealing temperature exceeds 1550 C.









1.3.2.2 Oxides on SiC

One of the principal benefits of silicon carbide is that it oxidizes to form a stable

surface layer of silicon dioxide releasing carbon dioxide in the process [31]. However,

the detailed properties of the oxide and in particular the interface between the SiC and the

SiO2 are significantly different from Si [31].

The oxidation rate is crystal-orientation dependent and is far slower on the Si face

than the C face, though, in general, much better properties are found for the oxide on the

Si face. Oxidation temperature are normally around 1100 C and unlike Si, a post

oxidation anneal in a hydrogen ambient is usually reported to have little effect at reducing

interface state density [32]. Interface state density on p-type material can be reduced to

levels below 1011 cm-2eV-1 in the lower half of the gap by the use of re-oxidation anneal

below 1000 C in a wet ambient [33]. However, this treatment does not seem to improve

the density of states in the upper half of the gap [34]. Interface state densities on the 4H

polytype rise very rapidly towards the conduction band edge, typically exceeding 1013

cm-2eV-1, whereas on 6H the density is an order of magnitude lower. This has been

ascribed to a carbon related acceptor that is located just below the conduction band edge

for 4H but within the conduction band for 6H [35]. These defects have not been

successfully removed by any standard surface treatment or post oxidation anneal.

The best N channel inversion mode MOSFETs fabricated on 6H show electron

mobilities of 100 cm2/Vs with a negative temperature coefficient, whereas 4H shows

mobilities of at best 25 cm2/Vs or lower [36, 37]. The fluctuation in potential resulting

from the charge in the interface states near the conduction band edge appear to be

responsible for these low values of mobility, particularly for 4H [38].









The breakdown and reliability properties of Si02 are crucial for all MOS devices,

but unfortunately the situation is not as favorable as it is for Si. The dielectric constant of

SiC is -10 whereas it is -3.9 for Si02, so any normal surface field will be -2.5 times

higher in the oxide than the SiC. Hence to gain the full benefit of the 2.5 MV/cm

breakdown field of SiC, the oxide must withstand 6.25 MV/cm, which is higher than is

reliably usable, even on silicon [39]. High field stressing of oxides has shown that oxides

grown in a wet ambient, breakdown at lower fields than dry grown oxides [40], and that

extrapolated time-dependent dielectric-breakdown lifetimes of 10 years can only be

obtained on n-type at fields less than 5 MV/cm at room temperature. Studies have shown

that the lifetime for oxides drops rapidly at elevated temperatures, with a vulnerability to

negative bias-stress instability [41]. Electron injection into the oxide is more efficient

than for Si due to the lower barrier [32]. In addition, the barrier for injection of holes is

much lower than in Si, which is unfortunate since holes are particularly damaging to

oxides [42].

1.3.2.3 Contacts

Ohmic contacts to both n and p-type material have been demonstrated for 4H-SiC

with specific contact resistivities of the order of 10-5 Qcm2 [43, 44]. To achieve this, it

has proved necessary to use rapid thermal annealing (RTA) of the wafer at temperatures

as high as 1400 C. The majority of contacts are based around Ni and Al (for n- and p-

type, respectively) although these are generally not stable at temperatures above 400 C.

To take advantage of the full potential for high temperature operation of 4H-SiC devices,

contacts that are thermally stable up to 600 C over long periods are required. Recent

work based on Al/Ni/W/Au contact structures on p-type 4H-SiC have shown longevity of

over 100 h at 600 C with a contact resistance of 10-3 Qcm2 [45]. Other works have









shown similar results using a variety of materials including TaC [46] and A1Si [47]

alloys.

Several groups have reported excellent quality room temperature Schottky diodes,

with the barrier height dependent on the metal chosen. This demonstrates that the fermi

level is not pinned at the surface of the 4H-SiC. Values of OB vary from 1.10 eV (Ti) to

1.73 eV (Au), which are markedly larger than those observed on Si or GaAs. This

increase in OB makes SiC an ideal material for high-voltage, high-temperature, low -

leakage current Schottky diodes. Surface preparation, prior to deposition of the Schottky

contact, has been shown to be a key factor in achieving good performance with sacrificial

thermal oxidation being the optimum technique [48].

1.3.2.4 Edge termination and passivation

Many techniques applied to Si devices are also applicable to SiC. For example,

field plates [49, 50], guard rings and junction termination extensions [51, 52] have all

been used to good effect. Another simple technique involves the implantation of a high

dose of inert ions (either Ar [53] or B [54]). In this case the damage caused by the

implant causes a high resistivity region to be formed close to the surface, which acts in a

manner similar to semi-insulating poly crystalline silicon (SIPOS). The reverse leakage

performance of Ar and B implanted edge terminations may be improved by low

temperature (600 C) annealing [55]. Passivation of SiC surface is not trivial and has yet

to be fully understood. The effects of inadequate passivation have been observed in

microwave IMESFETs, where this leads to degradation of gain under CW operation [56].

Power switching devices, on the other hand, appear to perform well with conventional Si

passivation treatments such as polyamide.














CHAPTER 2
SILICON CARBIDE SCHOTTKY RECTIFIERS

Silicon carbide device technology has seen a tremendous development over last

five years. The technology once envisioned as a substitute to silicon technology has been

realized to practice. The feasibility of SiC devices has been shown for many different

types of devices, the development of a working production technology has started, yield,

reliability and costs now being the key issues. The high cost of SiC substrates at present

poses a major challenge for the technology to enter the device market. However, the

exceptional properties of the material promote a tremendous interest in some of the prime

applications like power electronics and high temperature sensor technology. The other

examples are in transportation and manufacturing sector where high efficiency needs

entails strong research efforts towards development of advanced power management and

control electronics. Central to this effort is the development of the solid-state devices

capable of delivering large currents and withstanding high voltages, without the need of

sophisticated cooling systems [56].

2.1 Experimental Characteristics of 4H-SiC Schottky Rectifiers

2.1.1 Introduction

SiC has shown outstanding potential for high power, high temperature electronics

in advanced communications and radar systems, sensor and control systems, utility power

switching and traction motor control [57-68]. The wide bandgap (3.25 eV for the 4H-poly

type), availability of doped substrates of either electrical conductivity type and high

thermal conductivity (up to 4.9 W/cmK) of SiC make it a preferred material for these









applications [69-77]. These characteristics result in significantly higher current densities,

operating temperatures and breakdown voltages than comparable Si devices and in lower

switching and on-state losses. There have been major advances in the bulk and epitaxial

growth of SiC in the past 5 years or so, along with improvements in device design and

fabrication. Those have resulted in very impressive performance from SiC power metal-

oxide semiconductor field effect transistors (MOSFETs), bipolar transistors and p-i-n and

Schottky rectifiers [78-93].

2.1.2 Experimental Methods

The starting substrates were n+ (1019 cm-3) 4H-SiC. Approximately 10jtm of lightly

n-type (n 5 x 1015 cm-3) 4H-SiC was grown on these substrate by vapor phase epitaxy,

followed by 500 A of thermal Si02. A full-area back contact of e-beam deposited Ni

annealed at 970 OC was used for contact to the substrate. Si02 layers deposited by plasma

enhanced chemical vapor deposition (PECVD) were employed as a part of the dielectric

overlap edge termination [94]. Holes were opened in the SiO2 stack by a combination of

dry and wet etching and a Schottky contact of e-beam evaporated (1000 A thick) Ni

patterned by lift-off. The contact diameter ranged from 100-1000 |tm. A schematic of the

completed structure and an optical micrograph of a processed 2 inch diameter wafer are

shown in Figure 2-1.

A key feature of achieving good Schottky performance is the quality of the ohmic

contacts. The contact resistivity -1.2 x 10-6 Qcm2 was obtained from transmission line

measurements of the Ni backside ohmic for 970 OC, 3 min anneals under a flowing N2

ambient in a Heatpulse 610 furnace. This is slightly higher than reported for annealed Ni













































Figure 2-1 Cross-section of SiC Schottky rectifiers (top) and photograph of completed 2
inch diameter wafer (bottom).


Epi n- SiC
(t1,=10 =m)
Substrate n' 4WSiC
(n+= le19)
Ni









on 6H-SiC substrates (7 x 10-7 Qcm2) [95], which have a slightly smaller bandgap (3.02

eV) than the 4H-polytype used here.

2.1.3 Results and Discussion

Figure 2-2 shows some reverse current-voltage (I-V) characteristics measured at

250C schottky contact diameter. Note that the reverse breakdown voltage decreases

significantly with increasing contact size. This is due to the increased probability of

having crystal defects in the active region of the device as the area is increased. The

production of SiC substrates with large, defect free areas still represents the biggest

challenge to widespread use of power rectifiers in this materials system. Note also the

very low reverse leakage currents in the rectifiers at biases below breakdown. In this bias

regime, the reverse current was proportional to the length of the rectifying contact

perimeter, suggesting surface contributions were the most important in this voltage range.

At biases closer to breakdown, the current was proportional to the area of the rectifying

contact. Under these conditions, the main contribution to the reverse current is from

rectifiers with a 7000 A thick SiO2 dielectric overlap layer, as a function of the under this

contact, i.e. from the bulk of the material.

A more detailed view of the reverse I-V characteristic from a 154 rtm diameter

rectifier is shown in Figure 2-3 (top), along with a map of VB values measured over a

quarter of a 2-inch diameter wafer (bottom). The overall shape of the I-V curve is similar

to past reports for Ni/4H-SiC rectifiers [90], in which it was concluded that the origin of

the reverse current was due to a combination of both thermionic field emission and field

emission. Note that yield of rectifiers with VB > 750 V is over 50 % and with VB > 600 V

is over 75 %.







18






10-4
S1- 100 tm diode
-0- 500 rm diode

106 \ 1000 tm diode



10-8
S10 -
--S--A.-






10-12
-800 -600 -400 -200 0
Bias Voltage (V)



Figure 2-2 Reverse I-V characteristics at 25 OC from Schottky rectifiers with different
contact diameters.














10'




10 SiC Diode
S5000 A SiO2
108 Dia: 154 gm


10


1010


10 11
-1000 -800 -600 -400 -200 0
Bias Voltage (V)


Figure 2-3 Reverse I-V characteristic from 154 |tm diameter rectifier (top) and map of
reverse breakdown voltage from a quarter of a 2 inch diameter wafer
(bottom).









Simulations of the rectifier performance were carried out using the MEDICITM

code. The performance of the structure of Figure 2-1 (top) was simulated for different

oxide thicknesses and a metal overlap of 10 |tm onto this oxide, as used experimentally.

Figure 2-4 shows a comparison of the experimental and theoretical results. Note that the

VB values obtained on the real devices are roughly half of that expected from the

simulations. The later do not take into account the presence of crystalline defects,

including micropipes in the SiC. This difference emphasizes the importance of crystal

quality in determining the performance of the rectifiers.

The reverse breakdown voltage is related to the epilayer doping (ND) through the

relationship [96]:


VB E-
2e ND

where s is the dielectric constant of SiC, Ec the critical field for breakdown and e the

electronic charge. Figure 2-5 shows our experimental data point for VB at a epilayer

doping concentration of 5 x 1015 cm-3 for 10 |tm thick layers, along with the expected

results from the simulations with different carrier concentrations. Note that there is not

much improvement expected for decreasing the carrier concentration below 5 x 1015 cm-3

since the layer will be depleted under those conditions.

Our experimental devices still have performance limited by material defect density

rather than purity. The VB values obtained in the simulations were found to increase

linearly with Si02 thickness from 2500 A to 17500 A and saturated thereafter.














-700


-800

-900

-1000

-1100

-1200-


-1300


-1400 F-


-1500


SiC Schottky Diodes

Doping n:5e15 cm3
t .:10 pm
ept





-U- Experimental results
-- Simulated results
I I I I I I I I I


2000 3000


4000 5000 6000 7000 8000 9000 10000 11000


Dielectric thickness (A)




Figure 2-4 Comparison of breakdown voltage obtained from simulations and the
fabricated devices.












However, thicknesses above -5000 A are difficult to achieve in practice because of

the long deposition times. The forward current characteristics of a Schottky rectifier are

dominated by the barrier height of the metal on the semiconductor and by series

resistance concentrations. The barrier height (4B) for Ni on 4H-SiC is -1.3eV [90]. The

forward voltage drop, VF, is related to the barrier height and on-state resistance, RON,

through the relation:

nkT J
VF nkTn(+ RoFN.JF
e A T

where n is the diode ideality factor, k is Boltzmann's constant, T the absolute

temperature, e the electric charge, JF the forward current density and A* is Richardson's

constant. The value of RON is a function of epilayer thickness and doping. Figure 2-6

shows forward I-V characteristics in both linear (top) and log (bottom) form. A maximum

forward current of 2 A was obtained from these devices, and was limited by our

experimental system. The turn-on voltage was -1.82 V at a forward current density of

100 Acm-2 and the on-state resistance was -0.36 mQcm2. The diode ideality was 1.13 at

25 OC, which is comparable to past results for Ni metallization [90].

To place our results in context, Figure 2-7 shows a theoretical plot of specific on-

resistance as a function of VB for Si, 4H-SiC and 6H-SiC, along with some experimental

data. The lines are derived from the relation [84-96]:

4Vs
RON -
,U E EC














U U


U

S
U


U


SiC Schottky Diode
t .=10 pm
tepli Om

tdielectricl1.0 pm


-m- Simulation Results
--- Experimental Result


Carrier Concentration (cm3)


Figure 2-5 Calculated reverse breakdown voltage as a function of epilayer doping.


'"" '""" '""" '""'













2.5

SiC Schottky Diodes
2.0 SiO2: 7000 A
Diode Dia.: 1000 pm

1.5


1.0 -


0.5


0.0 I I
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
Bias Voltage (V)









102






,10

I. / SiC Schottky Diodes
t.I SiO : 7000A
2
/ Diode Dia.: 1000 ftm


100 I -1- I I I
0.0 0.5 1.0 1.5 2.0 2.5 3.0
Bias Voltage (V)


Figure 2-6 Forward I-V characteristics in linear (top) and log (bottom) scales.










10l


Diode Rectifiers

5 10-
E -
Si / ABB '98
o -7
S10-2 Cncinnati'97 Pur4de '9
Cree 02l *I Pyfdue 8
/ F lIps '97 v1
S/ Simern- s 97 ot .Yniv'95
/ NCSU'95 / re 2
6H-SiC
S13 ----------- -------
4H-SiC Rp 98
-- UF/Sterling



101 102 103 104
Breakdown Voltage (V)
Figure 2-7 Specific on-resistance as a function of breakdown voltage.









where [t is the electron mobility in the epilayer. Note that the SiC rectifiers are expected

to have either much lower on-resistance at a given breakdown voltage than Si, or

equivalently a much larger breakdown voltage for a given on-resistance. Our results are

close to the theoretical minimum on-resistance for the VB'S obtained and indicate that the

device processing produces no major degradation in the electrical properties of the SiC.

2.2 Ni Contacts to N-Type 4H-SiC

Superior electrical and thermal properties of SiC make it promising material for

high temperature and high frequency electronics. However, it is widely appreciated that

the optimal performance of microwave devices is related to the quality of ohmic contacts.

Low contact resistance and high temperature stability are required [97]. Nickel is one of

the most popular materials used in advanced aerospace systems [98] and good nickel

ohmic contacts on n-SiC have been already demonstrated [99]. The popularity of Ni

contacts to SiC is due to its reproducibly low contact resistance and good temperature

stability [100]. In our study, we could see no appreciable degradation of the contacts up

to a temperature of 225 oC.

Annealing the contacts produces low ohmic contact resistivities on highly doped

substrates. For n-type SiC, low contact resistivities (<10-5 Q cm-2) have been reported in

the past [101]. We obtained a contact resistivity of 1.2 x 10-6 Q cm2, in our studies. A

study on the interfacial reactions between Ni and n-type SiC suggests that silicides such

as Ni31Si12 and Ni2Si are produced at an intermediate temperature range of 500-600 C

[102]. X-ray diffraction (XRD) and Rutherford backscattering spectroscopy (RBS)

showed that the interfacial reaction forming Ni31Si12 and Ni2Si began at 500 oC [103], and

the Ni2Si phase remained up to 900 oC. The ohmic contact formation was thought to be









due to the formation of Ni2Si phase, even the Ni2Si formed at a temperature as low as 600

C [101]. It is essential to optimize the annealing conditions to obtain contacts with least

contact resistivity. In our studies we have studied the effect of temperature, ambience and

time on the contact resistivity of Ni-SiC ohmic contacts. Figure 2-8 shows the effect of

annealing temperature (top-left), time (top-right) and annealing ambient (bottom) on the

contact resistivity obtained from transmission line measurements of the Ni backside

ohmic. A minimum value of -1.2 x 10-6 Qcm2 was achieved for 970 OC, 3 min anneals

under a flowing N2 ambient in a Heatpulse 610 furnace. This is slightly higher than

reported for annealed Ni on 6H-SiC substrates (7 x 10-7 Qcm2) [95], which have a

slightly smaller bandgap (3.02 eV) than the 4H-polytype used here.

2.3 Stability of SiC Schottky Rectifiers to Rapid Thermal Annealing

2.3.1. Introduction

There is currently a lot of interest in the application of SiC power rectifiers to high

current switching in the >30A range because of the high thermal conductively

(-4.9W/cm-K) and large bandgap relative to Si [104-112]. The advantage in materials

properties mean that on-state resistance, RON, for a 4H-SiC rectifier can be over 100

times smaller than that of a Si rectifier at the same breakdown voltage, VB, since RON is

given by 4VB/teEc3 (where [t is the electron mobility, F the permittivity, and Ec the

critical field for breakdown) [87].

There are numerous situations in which high temperature annealing might be

employed in the fabrication of SiC rectifiers, including activation of implanted dopants or

reduction of reverse leakage current in self-aligned Ar -implanted edge terminations
















2 3x10





1 9x106





1 5x10s



















4"
S1 x10














1 8x106




1 4x10s




1 Ox0O











1 7x10s




S1 5x106




1 3x10




1 1x10


Figure 2-8 Contact resistivity of Ni ohmic contact on n 4H-SiC as a function of

annealing temperature (top), annealing time (center) and annealing ambient

(bottom).


Anneal time: 3 min
Anneal gas: N2














950 960 970 980 990 1000
Temperature (oC)







Anneal temperature: 970 C
Anneal gas: N2














20 25 30 35 40
Time (min)







Anneal time:3 min
Anneal temp:970 oC














N2 Vacuum Ar
Anneal Gas









[113-115]. In the latter case, Ar implantation is used to create a high-resistance region

around the contact periphery in order to spread the electric field distribution and avoid

breakdown due to field crowding. While this process does increase reverse breakdown

voltage, it also increases leakage current. Several studies have examined the effect of

anneals up to 7000C for reducing this leakage current [114].

In this study we report the stability of 4H-SiC Schottky power rectifiers to rapid

thermal annealing treatments in the temperature range 700 1100C with the contacts

already in place. This is of interest for defining the stability of devices in which the Ni

rectifying contact is used as a self-aligned mask for implant edge termination. The

performance of the rectifiers is found to show a general improvement for anneals

<10000C.

2.3.2 Experimental Methods

The Schottky contacts were fabricated on the wafer as described in section 2.1.2.

The annealing for these contacts was performed in a Heatpulse 610T System under

flowing N2 ambient at temperatures of 700 1100 oC for 60 240 sec. The current-

voltage (I-V) characteristics were measured before and after this procedure. Prior to

annealing, the typical on-state resistance was 0.8 mQ-cm2, the reverse breakdown voltage

-2
(VB) -450V and the forward turn-on voltage was 2.5V at 100Acm-2

2.3.3 Results and Discussion

The reverse bias characteristics showed a general increase in current for anneals <

1000C, while the forward I-V characteristics also shoed significant changes under these

conditions, as shown in Figure 2-9. Note the decrease in turn-on voltage as the anneal

temperature increase. The forward current density, JF, can be expressed as:









JF =AT2 eexp q -1
^ UkT }nkT)

where A** is the Richardson's constant for SiC, k is Boltzmann's constant, e is the

electric charge, q B is the barrier height for Ni on 4H-SiC, V is the applied voltage, n is

the ideality factor, and T is the measurement temperature. The data in Figure 2-9 are

consistent with a reduction in O B. On 6H-SiC, Ni was observed to form Ni2Si at

temperature beginning at -600C [116], eventually leading to ohmic behavior. In lower

temperature anneals, Ni/6H-SiC diodes showed a reduction of leakage current through

removal of low-barrier secondary diodes in parallel to the primary diode [113]. Our

measure O Bwas ~1.4eV prior to annealing, which is consistent with past reports [116-

118]. It is likely that the reduced turn-on voltage observed in Figure 2-9 is the result of

silicide formation, as suggested previously [114]. The on/off current ratio was 1.5 x 105 at

3V/-450V in control samples and increased by a maximum of -20% after optimum

annealing at 700C. The forward current characteristics of a SiC Schottky rectifier are

dominated by the Ni barrier height and by series resistance contributions and is related to

B and RON through the relationship:

VF lnIn 2 +noB + RNJF
i nk **T2J, J- _2
e AT

With annealing, the value of O B decreased as described above, but RON showed

improvement (Figure 2-10). Once again the annealing durations had to be kept<120 sec at

the highest temperature in order to retain some acceptable rectifying behavior of the top

Ni contact.









Figure 2-11 shows the effect of annealing time at 900C on the forward I-V

characteristics (top), VF (center), and RON (bottom). Note that the largest anneal (240 sec)

produced essentially ohmic behavior. Examples of the effect of the annealing duration on

the top Ni contact morphology are shown in the optical micrograph of Figure 2-12. It is

clear that there is already a strong metallurgical reaction between the Ni and the SiC at

1000C for 60 sec. To obtain the lowest ohmic contact resistivity for annealed Ni contacts

requires -3 min at this temperature, and after 60 sec the contact still retains some

rectifying behavior. Note also the dark appearance of the contact after annealing. Auger

Electron Spectroscopy showed high levels of carbon present in these reacted contacts,

suggesting the reaction can be written as:

2Ni + SiC Ni2Si + C

Figure 2-13 shows the measurement temperature dependence of the change in VF

(top) and RON (bottom) for samples annealed at 900C. There is little significant change

in VF while the improvement in RON decrease with increasing temperature as current

transport over the barrier. Arrhenius plots of the forward current indicated a barrier height

of ~1.14 eV after the 900C anneal, consistent with the observed increase in current after

annealing. The intersection of the lines with the temperature axis indicated a constant

electrically active contact area equivalent to -9% of the physical contact area. This is

consistent with the expected behavior for thermionic emission still being the dominant

transport mechanism after 900C annealing.









2.3.4 Summary and Conclusions

The main points of our study may be summarized as follows:

* High voltage Ni/4H-SiC Schottky rectifiers using a simple SiO2/metal edge
termination method show near-ideal forward characteristics (ideality factor, on-
state resistance).

* The reverse breakdown voltage is approximately half that expected from a simple
model simulation and decreases as the device area is increased.

* The reverse current appears to be dominated by surface combinations at low bias
and by bulk combinations near breakdown.

* Both thermionic field emission and field emission mechanisms appear to be
present. The yield of 750 V, 2 A devices was above 50 %.

* For high temperature anneals, the Ni rectifying contact becomes ohmic, with severe
degradation of the contact morphology. The forward current characteristics are still
dominated by thermionic emission until the onset of ohmic behavior.

* The benefits of annealing at moderate temperature include a decrease in forward
turn-on voltage and an increase in the on/off current ratio.

* The reasons for those improvements may include annealing of defects or reduction
of interfacial oxides between the Ni rectifying contact and the SiC.

* The use of more thermally stable rectifying contact metallization would allow
higher annealing temperatures and possibly more improvement in device
performance.












0.030 -


0.025


0.020


0.015


0.010


0.005


0.000 -
0.0


2.0 2.5 3.0


Figure 2-9 Forward I-V characteristics as a function of anneal temperature for 60 sec
anneals.


0.5 1.0 1.5

Bias Voltage (V)














4.0


3.5 60 sec

anneals
3.0


2.5


0 2.0


1 1.5


1.0


0.5


0.0
700 750 800 850 900 950 1000 1050 1100

Temperature (oC)


Figure 2-10 Change in RON as a function of anneal temperature for 60 sec anneals.
















0035


0030

240 secs 60 secs Control
0025


0 020


0015


0010


0005


000
00 05 10 15 20 25 30 35 40
Bias Voltage (V)




-1 1

-12

-13

14

-15 -










-20
40 60 80 100 120 140 160 180 200 220 240 260
Anneal Time (Seconds)




225

2 00

175

1 50

1 25

Z1 00

<0 75

060
0 50

0 25

000 i i i
40 60 80 100 120 140 160 180 200 220 240 260
Anneal Time (Seconds)





Figure 2-11 Forward I-V characteristics (top) and change in VF (center) and RoN (bottom)

as a function of anneal time at 900TC.




















































Figure 2-12 Optical micrographs of Ni contacts before (top) and after annealing at
1000C for 60 sec (center) or 120 sec (bottom).





































20 40 60 80 100 120 140 160

Measurement Temperature (C)


I I I I I I 1
20 40 60 80 100 120 14(

Measurement Temperature (C)


Figure 2-13 Change in VF (top) and RON (bottom) as a function of measurement
temperature after 900C, 60 sec anneals.


-0.30


-0.35


-0.40


-0.45


-0.50


I *














CHAPTER 3
IRRADIATION AND PASSIVATION EFFECTS ON 4H-SIC RECTIFERS

3.1 High Energy Proton Irradiation Effects on SiC Schottky Rectifiers

3.1.1 Introduction

SiC power rectifiers with reverse breakdown voltages in the 300-1200 V range are

of current interest for a number of applications, including traction motor control, sensors

and advanced communications (including broad-band satellite transmission) and radar

systems [105-108]. In particular, 4H-SiC is attractive because of its larger bandgap and

higher and more anisotropic mobility relative to other SiC polytypes [119]. The 4H-SiC

has numerous advantages over Si for power device applications, including much larger

critical breakdown field, lower on-state resistance at a given voltage and higher thermal

conductivity. While a significant amount of work has been reported on ion-beam induced

damage and annealing in SiC [120-123], much less is known about radiation damage in

SiC-based devices. Recent results have shown that properly passivated SiC metal-oxide

semiconductor field effect transistors have good tolerance to MRad doses of 60Co y-rays

[124]. One would expect SiC to be relatively resistant to radiation damage compared to

materials such as GaAs based on the empirical observation that the displacement

threshold energy for atoms in semiconductor is inversely proportional to lattice constant

[125] (a = 0.31 nm, c = 1.03 nm for 4H-SiC compared to a = 0.5653 nm for the zinc

blende GaAs).

There is particular interest in the response of SiC rectifiers to high-energy proton

irradiation because of the potential applications in satellites. In this work we reported on









the effects of 40 MeV proton irradiation of 4H-SiC rectifiers, at doses corresponding to

more than 10 years in low-earth orbit. The reverse leakage current increases by

approximately a factor of two under these conditions, while ideality factor, on-state

resistance and forward turn-on voltage all increase by < 75%. These results are consistent

with creation of electron traps by protons that traverse the active region of the rectifiers.

3.1.2 Experimental Methods

The Schottky rectifiers with diameter of 154 tm were fabricated in the same way

as described in section 2.1.2. The devices were irradiated at the Texas A&M cyclotron

with 40 MeV protons at fluences from 5 x 107-5 x 109 cm-2, with the highest fluence

being comparable to at least 10 years exposure in low earth orbit. The projected range of

the protons is >50 |tm in SiC. The irradiated devices were measured approximately 2

days after exposure.

3.1.3 Results and Discussion

Figure 3-1 (top) shows reverse current voltage (I-V) characteristics before and

after a proton fluence of 5 x 109 cm-2. The basic shape of the characteristic is unchanged

and shows only an increase in the magnitude of the reverse leakage current. Figure 3-1

(bottom) shows the percentage increase in reverse current measured at -250V, as a

function of the proton fluence. We suggest the creation of generation-recombination

centers related to atomic displacements created by the incoming protons is the main

mechanism for the increased leakage current, based on both this data and what follows

later in this section.

The forward I-V characteristics before and after proton irradiation at fluence of 5

x 109 cm-2 are shown at the top of Figure 3-2. There is an increase in low-bias (< 1V)









current consistent with the introduction of recombination centers associated with proton-

induced displacement damage. At higher forward voltages, the current is decreased.

Figure 3-2 (bottom) shows the magnitude of the decrease (measured at a forward voltage

of 2 V) as a function of the proton fluence. Note the percentage decrease in current is ~

6% after a fluence corresponding to more than a year in low earth orbit and 42% after

more than 10 years. The Schottky barrier height was extracted from the relationship:

2 e eV
JF =A T exp(- k")[(e)- 1]
kT nkT

where JF is the forward current density at voltage V, A** is the Richardson's constant for

4H-SiC, T the absolute measurement temperature, e the electronic charge, QB the barrier

height for Ni on 4H-SiC, n the ideality factor and k the Boltzmann's constant. The

extracted values were 1.32 0.06 eV for both control and irradiated rectifiers, which is

consistent with past reports for Ni on 4H-SiC [54].

The forward voltage drop for a Schottky rectifier, VF, is related to the barrier

height and on-state resistance, RON, through:

nkT JF
VF n(-- ) + + RoNJF
e AT

VF is usually defined as the voltage at which JF is 100 A.cm-2. Figure 3-3 shows

the increases in VF, n and RON as a function of proton fluence. The values in the

unirradiated control rectifiers were VF = 2.4V, n = 1.15 and RON = 3.7 x 10-4 Qcm2. In

each case, the values of these parameters increase with increasing proton fluence. The

results are consistent with a net reduction in carrier density in the depletion region of the

rectifiers through the introduction of traps and recombination centers associated with

proton damage.









3.2 Influence of PECVD of SiO2 Passivation Layers on 4H-SiC Schottky Rectifiers

3.2.1 Introduction

There is great current interest in the development of manufacturable SiC power

rectifiers in the 300-1000 V, 25-50 A range for use in traction motor control, sensor and

control systems and the drive train of hybrid electrical vehicles [119-126]. Numerous

reports have shown the potential of both SiC Schottky and p-i-n rectifiers for these

applications [84, 87]. Record breakdown voltage (12.3 kV) and forward current (130 A)

results have been achieved for these devices [87]. The main focus now is to understand

the effect of crystal growth and device processing steps in the performance of SiC

rectifiers, in an attempt to push the devices into a robust manufacturing mode. One of the

important fabrication steps is the deposition of the thick dielectric layers for use as

surface passivation and as a part of the metal overlap edge termination. These layers are

generally deposited by plasma enhanced chemical vapor deposition (PECVD) using

SiH4-based chemistries and therefore the surface of the SiC is subject to both energetic

ion bombardment and a high flux of atomic hydrogen. The former can create surface and

bulk traps that degrade the electrical properties of semiconductors while the later can etch

the surface through the reaction:

SiC + 8H SiH4 + CH4

as well as passivate dopants in the near-surface region. Both donors (D) and acceptors

(A) can be passivated through the formation of neutral complexes with hydrogen,

according to following mechanisms:

D++ H- (DH)0

A- + H+ (DH)0












10-4

10-5

10-6

10-7

10-8

10-9

10-10

10-11







103


-500 -400 -300 -200
Bias Voltage (V)


8 9


108 109

Proton Dose (cm-2)


-100 0


1010


Figure 3-1 Reverse I-V characteristics from 4H-SiC Schottky rectifiers before and after
proton irradiation at a dose of 5 x 109 cm-2 (top) and percentage increase in
reverse leakage at -250V (bottom).


SiC Schottky diode
SiO2:7000 A
Dia.: 1000 gim












100






10-1






10-2


0.0 0.5 1.0 1.5 2.0 2.5 3.0
Bias Voltage (V)


SiC Schottky diode
40 SiO2:7000 A
Dia.: 1000 pm


7 108 109 1

Proton Dose (cm-2)


Figure 3-2 Forward I-V characteristics from 4H-SiC Schottky rectifiers before and after
proton irradiation at a dose of 5 x 109 cm-2 (top) and percentage decrease in
forward current at 2V (bottom).


SiC Schottky diode
SiO2:7000 A
Dia.: 1000 pm
dose: 5x109 p/cm2


- pre-irradiated
- post-irradiated


I I I























0.5


0.0 L
107


108 109
Proton Dose (cm 2)


3 -



2



1-




107 108 109 10
Proton Dose (cm )


1.0






t 0.5
o


0.0
107


108 109
Proton Dose (cm )


Figure 3-3 Numerical change in values of n (top), RON (center) and VF (bottom) as a
function of proton fluence.


I I


"""" """" """'









In this study we report on the effects of PECVD SiO2 layers on the performance

ofNi/4H-SiC Schottky rectifiers. We also identify optimum deposition conditions under

which there is minimal degradation of the device performance.

3.2.2 Experimental Methods

The Schottky rectifiers with a diameter of 154 [tm were fabricated in the same

way as described in section 2.1.2. PECVD of thin (-200 A) SiO2 films were performed at

a substrate temperature of 250 OC using 2% SiH4 and N20 at a total flow rate of 600

sccm. The process pressure was raised from 600-900 mTorr, the N20 percentage of the

total flow rate from 5 to 50% and the rf power (13.56 MHz) varied from 25-400W. The

SiO2 layers were thin enough that we could probe through them to the underlying

contacts, allowing us to measure the current voltage (I-V) characteristics without having

to remove the SiO2 layers. Prior to SiO2 deposition, the rectifiers displayed reverse

breakdown voltage (VB) of 770 V, on-state resistance (RON) of 4.73 mQ.cm2 and forward

-2
turn-on voltage (VF) of 2.06 V at 100 A.cm-2

3.2.3 Results and Discussion

Figure 3-4 shows the effect on VB of plasma power (top), pressure (center) and

N20 percentage (bottom) during the SiO2 deposition. There is no significant change in VB

at low powers, while at higher powers (>250 W), it increases. Since VB is inversely

dependent on the doping in the depletion region (ND), according to the relation:



B 2e ND

where e is the permittivity of SiC and e the electronic charge, this suggests a decrease in

doping in the unprotected region around the rectifying contact (and possibly under, since

atomic hydrogen ion diffuse in an isotropic fashion). This effect was largest at the lowest









process pressure, as seen in the center part of Figure 3-4. The bottom part of the figure

shows that VB actually decreased under SiH4-rich deposition conditions. Therefore, we

suggest that the main effect of the hydrogen flux is to decrease VB through creation of

surface states, whereas the effects of plasma power and pressure are mainly due to

creation of deep traps by energetic ion bombardment. In this later case, the effective

doping around the contact is decreased and this leads to an increase in breakdown

voltage.

The forward turn-on voltage was less sensitive to changes in the SiO2 deposition

parameters than was VB. This is expected, since VF is controlled by the barrier height of

Ni contact (1.3 eV) and by series resistance contributions, especially from contact

resistance. Since the area under the Schottky contact is protected by the overlying metal

and the rear ohmic contact is not exposed to the plasma, we would expect VF to be less

sensitive to the process conditions. Figure 3-5 shows the effect on VF of both plasma

power (top) and process pressure (bottom), with the changes in each case being less than

S10%.

The on-state resistance is related to the critical field for breakdown Ec and the

reverse breakdown voltage through the relation [96]:

2
4V;
RON /ICEC


where |t is the electron mobility in the 4H-SiC epi-layer. Figure 3-6 shows the effect of

plasma power (top), N20 percentage (center) and process pressure (bottom) on RON. The

trends mirror those seen for VB under the same conditions, with the increases being

greatest at the highest powers, highest SiH4 content in the plasma and the lowest process

pressure.









3.3 Effect on 4H-SiC Schottky Rectifier of Ar Discharges Generated in a Planar
Inductively Coupled Plasma Source

3.3.1 Introduction

SiC-based rectifiers are generating tremendous current interest for a broad range

of applications, including broad band satellite transmission systems, advanced radar, high

temperature sensors and traction motor control [119-127]. Most attention has been

focused on the 4H-SiC polytype because of its larger bandgap (3.25 eV) and higher

mobility relative to the other polytypes [70]. Within the two basic classes of rectifiers,

Schottky devices have the lowest on-state voltages and highest switching speeds; while p-

i-n diodes have the higher reverse breakdown voltage and lower reverse leakage current

[87]. Compared to Si rectifiers, SiC devices have on-state resistances approximately a

hundred times lower or equivalently much larger breakdown voltage at the same on-state

resistance [96]. While very high forward currents (up to 130 A) and breakdown voltages

(4.9 kV) have been achieved for 4H-SiC Schottky rectifiers [87], a lot of recent attention

has been focused on fabrication and materials technology for diodes in the 300-1000 V

range [105]. Plasma processing is required both for dry etching of mesas and deposition

of dielectrics for surface passivation and metal overlap edge termination. However little

has been reported on the effects of plasma exposure on the electrical performance of 4H-

SiC rectifiers.

In this study, we describe the effect of Inductively Coupled Plasma (ICP) Ar

discharges generated in a novel plasma source configuration, on the electrical properties















DU
SiC Schottky diodes


30 -

20 -

10

0

-10

-20

-30
0 100 200 300 400
Power (W)






--- SiC Schottky diodes
40



20



0



-20 i
600 700 800 900
Pressure (mTorr)




0
SiC Schottky diodes

-20






-60


-80


inn


0 10 20 30
% N20


40 50


Figure 3-4 Percentage change in VB as a function of plasma power (top), process pressure
(center) and N20 content (bottom) during SiO2 deposition.







49





12 SiC Schottky diodes


10




6


4


2 I
0 100 200 300 400
Power (W)





10 SiC Schottky diodes


8


6


4

2 -

600 700 800 900
Pressure (mTorr)



Figure 3-5 Percentage change in VF as a function of plasma power (top) and process
pressure (bottom) during SiO2 deposition.
















25 SiC Schottky diodes


20


.5 15


1 10


5


0 *
0 100 200 300 400
Power (W)





10 SiC Schottky diodes


8


" 6


0 4


2


0 I
0 10 20 30 40 50
% NO






25 SiC Schottky diodes


20


.E 15


| 10


5


0
600 700 800 900
Pressure (mTorr)


Figure 3-6 Percentage change in RON as a function of plasma power (top), N20 content
(center) and process pressure (bottom) during SiO2 deposition.









ofNi/4H-SiC Schottky rectifiers. Both ion flux and ion energy are found to play a role in

the extent of the observed changes in reverse breakdown voltage (VB), on-state resistance

(RON), ideality factor (n) and forward turn-on voltage (VF).

3.3.2 Experimental Methods

The Schottky rectifiers with a contact diameter of 154 |tm were fabricated in the

same way as described in section 2.1.2. The as-fabricated rectifiers show RON of 4.4 mQ-

cm2, VF = 1.96 V and VB = -330 V.

The devices were exposed to pure Ar discharges in a planar coil geometry ICP

reactor, shown schematically in Figure 3-7. A key feature of this system is the ability to

operate at very low source power (20 W) as well as chuck power (5 W), which allows

one to design etch or deposition processes that end with very low ion energy and flux

conditions to minimize damage. The ICP source power (2 MHz) was varied from 100-

700 W, the rf chuck power (13.56 MHz) from 25-200 W and the process pressure from

10-40 mTorr. All exposures were 30 sec in duration and removed < 150 A of the surface

by sputtering. These plasma exposure simulate the ion bombardment received by the SiC

surface during processes such as dry etching of plasma enhanced chemical vapor

deposition and represent a worst-case scenario because there is no chemical etch

component that would remove some of the damaged surface region on a deposition of

dielectric to protect the surface from further ion bombardment.

3.3.3 Results and Discussion

Figure 3-8 shows the percentage changes in VB (top) and RON (bottom) as a

function of ICP source power (100 W) and process pressure (10 mTorr). The reverse

breakdown voltage decreases with increasing ion flux. Since the dc self-bias on the









sample electrode also decreases with increasing source power, this indicates that

increasing ion flux is a factor in degrading the breakdown voltage. Note however that the

changes are < 25% even at the highest source power employed. For SiC there is an

empirical relationship between VB and the doping in the epilayer, ND, which is given by

[96]:


V= 1.75 x 150 JD075

One of the expected effects of Ar plasma exposure on the exposed SiC around the

contact periphery is a decrease in ND through creation of deep trap states. However this

would lead to an increase in VB, which is the opposite of what is observed

experimentally. Therefore, we suggest that the main effect of the plasma exposure under

our conditions is an increase in surface defects that initiate breakdown at much lower

values than expected from the bulk doping of the SiC epilayer.

The on-state resistance can be written as [106]:

2
RON -


where [t is the electron mobility, e the permittivity of 4H-SiC and Ec the critical field for

breakdown. Since VB decreases with increasing source power, we would also expect a

decrease in RON unless Ec is also decreasing. Figure 3-8 (bottom) shows large increases

in RON as the source power is increased, which is consistent with the creation of surface

states that promote early breakdown of the damaged rectifier.

Figure 3-9 shows the percentage changes in VF (top, measured at 100 A.cm-2) and

ideality factor (bottom) as a function of source power at fixed rf power (100 W) and

pressure (100 mTorr). Note that the changes in both parameters are < 20% provided the









source power is left below 300 W, corresponding to a dc self-bias < -185 V. At higher

source powers both VF and n are severely degraded even though the self-bias under these

conditions is low (-41 V at 700 W source power). The average energy of the incident Ar+

ions is roughly given by the sum of the dc self-bias and the plasma potential

(approximately -25V in this system). Therefore even though the ion energy is low at high

source powers, the ion flux is sufficient to create significant degradation in the rectifier

characteristics. The forward voltage drop for a Schottky rectifiers is related to the barrier

height (4B) and RON through the relation [56]:

nkT J
VF-e n **TARoNJF

where k is Boltzmann's constant, T the absolute temperature, e the electronic charge, JF

the forward current density and A** is Richardson's constant for 4H-SiC. Therefore VF

can be degraded by increases in ideality factor and on-state resistance along with a

reduction in forward current due to introduction of trap states.

The changes in rectifier performance were also dependent on the applied rf chuck

power, or equivalently on ion energy. The changes in VB, VF, RON and n were < 20% for

VB and VF, < 30% for RON and < 40% for n over the range of these parameters that we

investigated.

Figure 3-10 shows the percentage changes in VB (top) and RON (bottom) as a

function of process pressure at fixed source power (300 W) and rf power (100 W). Note

that the decreases in both parameters are largest at the lowest pressure. This is consistent

with the fact that at low pressures the incident ions have a lower probability of collisions

with gas molecules or they traverse the sheath region and therefore they impact with their

full energy. A major advantage of the plasma ICP source is its ability to operate at higher









pressures (> 40 mTorr) relative to the more normal 10mTorr range of conventional

cylindrical coil sources. The ideality factor of the rectifiers was still more degraded at

higher pressures, while VF showed very small changes with pressure (Figure 3-11).

3.3.4 Summary and Conclusions

In conclusion 4H-SiC Schottky rectifiers show only modest changes in the

electrical characteristics after high-energy proton irradiation at fluences corresponding to

more than 10 years in low-earth orbit. These devices appear promising for both aerospace

and terrestrial applications where irradiation hardness is a prerequisite. The main

degradation mechanism appears to be creation of recombination centers and traps that

cause an increase in reverse leakage current, ideality factor and on-state resistance.

In another study, PECVD SiO2 layers were deposited on Ni/4H-SiC Schottky

rectifiers as a function of various plasma parameters. Both reverse breakdown voltage

and on-state resistance increase with increasing plasma power and SiH4 content through

creation of deep trap states and degradation of the SiC surface by reaction with atomic

hydrogen, respectively. The changes in forward turn-on voltage are less dramatic in each

case. Plasma conditions that utilize low-to-moderate powers, high process pressure and

low SiH4 percentages produce minimal changes in the rectifier performance.

4H-SiC power rectifiers were also exposed to Ar discharges in a plasma coil ICP

reactor as a function of various plasma parameters. The reverse breakdown voltage

decreases with increasing ion flux, which is controlled by the source power, whereas VF,

RON and n all increase under the same conditions. These results are consistent with

creation of surface states that promote reverse breakdown at lower applied voltages. The

same trends were observed with increasing of chuck power, which controls the incident

ion energy. The increases in VB and RON are minimized at high operating process






55


pressure, as expected since these conditions reduce the effectiveness of the ion

bombardment experienced during the plasma exposure. Damage during actual etch or

deposition process would be lower than produced during the Ar plasma exposure, but

clearly process conditions that utilize low ion energies and fluxes are desirable at the end

of an etch cycle or the beginning of a deposition cycle in order to minimize ion-induced

damage.











ICP (Inductively Coupled Plasma) Etching System
Frame Structure


Quartz

Reactor


Sample


Powered Electrode


He Cooling System


13.56 MHZ


Figure 3-7 Planar coil ICP reactor.



























-20 k-


I 0 I 0 I 0 I 0 I 0 I 7 I
100 200 300 400 500 600 700


500


400


300 i


200 *
u

100


0


ICP Power (W)


1400

1200

1000

800

600

400

200

0

-200


I I I I I I I
100 200 300 400 500 600 700

ICP Power (W)


Figure 3-8 Percentage change in VB (top) and RON (bottom) of 4H-SiC rectifiers as a
function of ICP source power.


SiC Schottky
RIE Power: 100 W
Pressure: 10 mTorr


SiC Schottky
RIE Power: 100 W
Pressure: 10 mTorr





I I I I I I I


I







58




250
SiC Schottky
200 RIE Power: 100 W
Pressure: 10 mTorr

150


100


S50


0

100 200 300 400 500 600 700
ICP Power (W)




80 ,
SiC Schottky
70 RIE Power: 100 W
S Pressure: 10 mTorr
60

S50

40

S30

20

10

100 200 300 400 500 600 700
ICP Power (W)


Figure 3-9 Percentage change in VF (top) and n (bottom) of 4H-SiC rectifiers as a
function of ICP source power.







59




5 .i.. -250


0 200


S-5 150


-10 100 a

SC/ SiC Schottky "
-15- RIE Power: 100 W 50
ICP Power: 300 W

-20 I I *-- I I I* 0
10 15 20 25 30 35 40
Pressure (mTorr)





32 -
32 SiC Schottky
30 \ RIE Power: 100 W
ICP Pwer: 300 W
28
e 26
S 24

S22
20
18
16
1 4 I I I I I I I
10 15 20 25 30 35 40
Pressure (mTorr)


Figure 3-10 Percentage change in VB (top) and RON (bottom) of 4H-SiC rectifiers as a
function of process pressure.













I I I I I I I
SiC Schottky
RIE Power: 100 W
ICP Pwer: 300 W













10 15 20 25 30 35 40
Pressure (mTorr)





SiC Schottky
RIE Power: 100 W
ICP Pwer: 300 W













10 15 20 25 30 35 40
Pressure (mTorr)


Figure 3-11 Percentage change in VF (top) and n (bottom)
function of process pressure.


of 4H-SiC rectifiers as a














CHAPTER 4
JUNCTION TERMINATION EXTENSION GEOMETRY OF SIC RECTIFIERS


It is possible to make etching of SiC (0001) facet only by plasma methods because

of its high chemical stability [128]. After such etching the surface of the semiconductor is

strongly damaged and is not electrically neutral. It is a reason for arising breakdown at

considerable lower voltages than the values implied by the impurity concentration in the

base. In most of the cases the breakdown characterized by surface leakage current (in the

order of 10-6-10- A) and was irreversible. For decreasing probability of surface like

breakdown it is necessary to make surface electrical field strength less than the strength

in bulk.

This can be achieved by profiling of mesa structure during etching [129] or by

diffusion o acceptor impurities into the surface region of the n-base [130]. Also electrical

activity of the surface can be decreased by special treatment like oxidation.

Another problem encountered for high reverse voltage is when the breakdown

occurs through the air along the surface of the mesa-structure. It has been shown that the

breakdown voltage increases when the measurements are done by placing the samples in

a dielectric liquid with a high value of electric breakdown field strength like fluorientTM

FC-77 [131]. For high temperature and high power devices if is necessary to cover mesa-

structure by the thermostable dielectric with critical electrical field strength more than

that of air.









4.1 Influence of Edge Termination Geometry on Performance of 4H-SiC P-i-N
Rectifiers

4.1.1 Introduction

4H-SiC has a larger bandgap (3.26 eV) compared to 6H-SiC (2.9 eV) and is

preferred for device applications because of its higher and more isotropic electron

mobility [70]. There is particular interest in the development of 4H-SiC power rectifiers

for high power electronics applications. The on-state resistance of SiC rectifiers can be

less than 1% that of Si rectifiers with the same breakdown voltage [106-108]. There are

two basic classes of rectifiers, namely Schottky and p-n junction. The former have lower

on-state voltages and higher switching speeds, while the latter have higher breakdown

voltage and lower reverse leakage current. Advanced designs such as junction barrier

Schottky (JBS) [109], dual metal trench [90] and merged p-i-n Schottky rectifiers (MPS)

[110] attempt to achieve the advantages of both types of rectifiers [84].

An empirically derived relationship of 4H-SiC shows that the critical electric field

for breakdown, Ec, is related to the doping level in the drift layer, ND, through [112]:

2.5 x106

1- 0.251og( )
106

or in terms of breakdown voltage, VB [104]:


V = 1.75x1015" 075
Excellent high current (230 A) and high voltage (5.5 kV) performance have been

reported for 4H-SiC p-i-n rectifiers. The performance of manufacturable devices in the

300-1000V range is of particular interest, especially the development of robust designs

and fabrication sequences. In this paper we report on the influence of edge termination

design on the performance of 4H-SiC p-i-n rectifiers with a simple epi layer structure.









4.1.2 Experimental Methods

Epi layers of 12 |tm of n-type (n-2xl015 cm-3) and 1 |tm of p-type (p~1017 cm-3)

4H-SiC were grown on n+ 4H-SiC substrates. A schematic of the completed rectifiers is

shown in Figure 4-1. A full-area back metallization of Ni annealed at 970 OC for 3 min

provided ohmic contact to the substrate. Mesas were formed by dry etching and thermal

SiO2 was used as the part of the dielectric overlap edge termination. The e-beam

evaporated Ni Schottky rectifiers had areas between 0.005-0.04 mm2. Passivation layers

of SiO2 (3000 A thick) were deposited by plasma enhanced chemical vapor deposition.

The adjustable parameters were the extent of metal overlap onto the mesa defining the

active device area and the length of this mesa. The mesa edge termination has been

demonstrated to provide excellent results on SiC p-n junctions provided the mesa surface

is properly passivated [90,94].

4.1.3 Results and Discussion

Figure 4-2 shows the forward I-V characteristics from one of the p-i-n rectifiers.

The I-V curve can be fitted to the standard relationship for p-n diodes, namely [107]:

I eV eV
J= JRO exp( ) + JDO exp( )
A nlkT n2kT

where A is the diode area, JRo and JDo are the saturation current densities for

recombination and diffusion current mechanisms, respectively, e is the electronic charge,

k is Boltzmann's constant, T the absolute temperature, V the applied bias and nl and n2

are the ideality factors for recombination and diffusion currents, respectively. A fit to the

data yields ni=1.97 and n2=1.10. Therefore, at lower current densities the transport is

dominated by recombination, whereas at higher forward voltages, the current is









dominated by the diffusion contribution. The forward turn-on voltage, VF, can be

represented as [84]:

kT n n
VF = n( )+ VM
e n,

where n. and n+ are the electron concentrations at the two end regions (p /n and n+/n), ni

the intrinsic electron concentration and VM the voltage drop across the drift region. The

measured VF is -4V, or roughly 0.74 V higher than the theoretical minimum. The best

on-state resistance was 15 mQcm2, with an on/off current ratio of 1.5x105 at 3/-450 V.

Figure 4-3 shows reverse I-V characteristics as a function of metal overlap distance

for fixed mesa extension distance of 60 itm. The reverse breakdown voltage shows a

trend of decreasing as the metal overlap distance increases. The mechanism is not clear at

the moment, but could be due to several factors, including breakdown in the thermal

oxide or extension of the depletion region to include defects under the mesa. The figure-

of-merit VB2/RON was a maximum of 13.5 MWcm-2 for these rectifiers.

Figure 4-4 shows the variation of VB with metal overlap distance for rectifiers with

different active areas. There is no significant effect of area over the range we employed,

but we would expect to observe a decrease in VB in very large rectifiers because of the

increased probability of having a defect in the active region [127].

Figure 4-5 shows reverse I-V characteristics from rectifiers with fixed area and

extent of metal overlap, as a function of mesa extension length. The VB values extracted

from this data are shown in Figure 4-6. Within experimental error and the uniformity of

VB on a given wafer (- 10%), there is no significant change in breakdown voltage as a

function of mesa extension length.









4.2 Effect of Contact Geometry on 4H-SiC Schottky Rectifiers with Junction
Termination Extension

4.2.1 Introduction

There is great current interest in SiC power Schottky rectifiers with reverse

breakdown voltages in the 500-1000V range for applications in traction motor control,

sensors and next generation communications and radar systems [125-132]. These devices

have significant advantages over Si diodes for these purposes, including lower switching

losses, higher operating temperatures and faster switching speeds [93, 96]. In addition the

high thermal conductivity of SiC (approximately three times that of Si) makes it ideal for

very high current applications. There is still a need to optimize the edge termination and

contact geometry for SiC rectifiers in order to find the most robust design for

manufacturing.

In this study we report on the effect of Schottky metal overlap and contact shape on

the reverse breakdown voltage of 4H-SiC rectifiers employing junction termination

extension (JTE). The JTE is used to reduce the electric field at the edges of the rectifiers.

It increases the breakdown voltage of the diode by reducing the electric field density

within the SiC near the edges of the rectifier. The p-doping of the JTE region counteracts

the bending of the depletion region around the edges of the anode. This effect spreads out

the electric field at the covers and edges of the depletion region.

4.2.2 Experimental Methods

The starting samples consisted of 1 rtm p-type 4H-SiC (p~1017 cm-3) on 12 rtm of

n-type 4H-SiC (n-2x1015 cm-3) on a n+ 4H-SiC substrate. A full area back contact of e-

beam deposited Ni annealed at 970 OC was used for contact to the substrate. Thermally













Passivating oxide


Figure 4-1 P-i-N rectifier.







67





1.4x10 -2


1.2x10-2 SiC P-i-N diode


1.0x10-2


8.0x10-3


6.0x10-3


Q 4.0x103-


2.0x10-3


0.0 ------

0 1 2 3 4 5

Bias Voltage (V)


Figure 4-2 Forward I-V characteristics of p-i-n rectifier.


I I


6 7





















10-7

10-8

109 SiC P-i-N diode

10 SiO2:3000 A p-SiC:1 pm
10- 2
Diode area: 0.01 mm
10o11 resq lengtI:60,pj
-700 -600 -500 -400 -300
Bias Voltage (V)


-200 -100


Figure 4-3 Reverse I-V characteristics from p-i-n rectifiers as a function of metal overlap
distance onto mesa.















600





0 400




S200
4 200


0 1 2 3 4 5 6

Metal Overlap (itm)


Figure 4-4 Variation of VB with metal overlap distance for p-i-n rectifiers.






























10-11


Si(
Si
Di
me







-U- mesa length: 60 pm
-*- mesa length: 40 pm
S-A- mesa length: 20 pm
-y- mesa length: 0 pm
, I I ,


C P-i-N diode
o2:3000 A p-SiC:1 pim
2
ode Area: 0.04 mm
;tal overlap: 4 jtm










I -


-500 -450 -400 -350 -300 -250 -200 -150 -100 -50 0
Bias Voltage (V)


Figure 4-5 Reverse I-V characteristics from p-i-n rectifiers as a function of mesa length.


I







71






450 -


400 -


S350


A 300


250
25 SiC P-i-N diode
200 SiO: 3000 A p-SiC: 1 gm
200 2
S Area: 0.04 mm Metal Overlap: 4 pm
150


1 0 0 i I I I iL
0 10 20 30 40 50 60

Mesa Length (gm)


Figure 4-6 Variation of VB with mesa length for p-i-n rectifiers with fixed metal overlap
distance of 4 tm.









grown SiO2 was on part of the dielectric overlap edge termination. Schottky contact holes

were etched into the SiO2 and through the p-SiC to expose the underlying n-SiC using

dry etching. Schottky contacts of e-beam evaporated Ni (area 0.04 0.64 mm2) were

patterned by lift-off. Three different contact shapes were employed, namely oval, circular

and square. Finally, Si02 layer deposited by plasma enhanced chemical vapor deposition

(PECVD) was used for passivation. A schematic of the completed device structure is

shown in Figure 4-7.

4.2.3 Results and Discussion

The forward turn-on voltage VF, defined as the forward bias at which the current

density is 100 A-cm2, was < 2V (Figure 4-8). The on-state resistance, RON, was

4.2mQcm2, which is close to the theoretical minimum for this structure of 4H-SiC. The

VF is -25% higher than the theoretical minimum for Ni Schottky diodes on 4H-SiC and

may reflect the presence of some residual dry etch damage or surface contamination.

Figure 4-9 shows the reverse current-voltage (I-V) characteristics from rectifiers

with 0.04 mm2 area, junction termination extension of 40 |tm and Ni metal overlap of 4

|tm in extent, as a function of contact shape. Note that the oval diodes have the highest

leakage at low bias (<250 V) and the lowest leakage at higher voltages. The reverse

breakdown increases in the sequence square
breakdown as the voltage at which the current reaches 1 |tm. It is not immediately clear

why the oval shape rectifiers have higher breakdown and more work is needed to

establish any relationship to defect density in the SiC samples or the role of surface states

around the contact periphery. The breakdown voltage of all rectifiers is still









approximately a factor of two lower than expected from the depletion layer doping and

thickness employed.

Figure 4-10 shows reverse I-V characteristics from rectifiers of different contact

area and fixed JTE (40 [tm) and extent of metal overlap (2 ptm). The reverse breakdown

voltage decreases with increasing contact area. This is most likely due to the increased

probability of having crystal defects in the active region of the rectifier as the area is

increased. The shape of the I-V curves is similar to those reported earlier for Ni/4H-SiC

rectifiers [90], where it was suggested that both thermionic field emission and field

emission mechanisms are present.

There was little effect of metal overlap extent on the reverse I-V characteristics, as

shown in Figure 4-11 for oval-shaped rectifiers with small area (0.04 mm2) and a JTE of

20 jtm. This is expected, since the field distribution in the depletion region is controlled

by the p-n junctions formed by the JTE regions on the drift region and not by the metal

overlying the p-regions. This gives considerable latitude in the patterning step for the

metal deposition. Similar results are shown in Figure 4-12 for rectifiers with a JTE length

of 40 ptm.

A summary of VB data as a function of metal overlap distance for different JTE

length is shown in Figure 4-13. While the number of samples is too small to draw

definitive conclusions, VB is maximized at shorter JTE lengths. We believe that

breakdown may actually be initiated in the p-layer under some conditions because of its

low doping level and therefore longer JTE regions have a higher probability of having

defects that promote breakdown. The diodes show an excellent on/off current ratio (1.5/-

500V) of> 4x105 at 25 OC.









4.3 Role of Device Area, Mesa Length and Metal Overlap Distance on Breakdown
Voltage of 4H-SiC P-i-N Rectifiers

4.3.1 Introduction

SiC p-i-n rectifiers have several advantages over their simpler Schottky

counterparts, most importantly, a higher reverse breakdown voltage (VB) and lower

reverse leakage current [104-110]. SiC p-i-n rectifiers are expected to offer significant

performance advantages over Si rectifiers for high voltage applications, rf base station

infrastructure, advanced avionics, combustion control electronics in automobiles and

solid-state power conditioning [133-135]. While very high VB and forward conduction

currents have been reported (8.6 kV [87]), the need is to design robust device structures

and processes for the 300-1200 V range. In particular, to realize a manufacturable SiC

rectifier process, it is necessary to understand the effect of various edge termination

geometries on the breakdown voltage characteristics so that VB most closely approaches

the expected parallel-plate value for the epi layer doping and thicknesses employed.

Moreover, it is necessary to carry out these investigations on 4H-SiC, which has larger

bandgap and higher and more isotropic mobility than the 6H-polytype [70].

In this study we report the results of an investigation into the role of device area,

mesa length and metal overlap distance on the breakdown voltage of 4H-SiC p-i-n

rectifiers fabricated on thick (-50 [tm) epi-layers under non-punch through drift region

conditions.

4.3.2 Experimental Methods

The starting substrates were n+ (1019 cm-3) 4H-SiC. Approximately 50 |tm of ntype

(-2 x 1015 cm-3) and 1 |tm of p-type (p 1017 cm-3) 4H-SiC was grown on these

substrates by vapor phase epitaxy, followed by -500 A of thermal Si02. A full area-back


































Figure 4-7 Schottky rectifier.







76







80 SiC Schottky Diode

(p-type edge termination)
c 0 SiO: 3000 A
S60 2
Diode Area: 0.64 mm2
y<
0 40



2 20



0 U


0.0 0.5 1.0

Bias Voltage (V)


Figure 4-8 Forward I-V characteristics of 0.64 mm2 rectifier.


I


I






77





10-4

10-5 SiC Schottky
area:0.04 mr
10-6 JTE:40 Lm

lo -7 metal overlay



10-9
10~9

-0 oval diode
-*-1 circular diode

.11 -A- square diode

-800 -700 -600 -500 -400 -300 -200
Bias Voltage (V)


Figure 4-9 Reverse I-V characteristics of 0.04 mm2 rectifiers with different contact shape.


-100 0







78






I* WAI

10_5 SiC Schottky Diodes
oval

10-6 JTE:40 gm
Metal overlap: 2 im

10 A

-8


109
0- \n n ,
S-- area: 0.04 mm

S-*-area: 0.16 mm
210 2 -\ -
1 -A-area: 0.64 mm2

10-11
10 I I -
-700 -600 -500 -400 -300 -200 -100 0

Bias Voltage (V)


Figure 4-10 Reverse I-V characteristics of rectifiers with oval shaped contacts of different
area.






79







10-5
SiC Schottky Diodes
10:6 (Oval)
area:0.04 mm2
10-7 JTE:20 pm

10 -

10-9
-u- metal overlap: 6 tLm
11 -10"metal overlap: 4 tm
-A- metal overlap: 2 |tm
1011
-800 -700 -600 -500 -400 -300 -200 -100 0
Bias Voltage (V)



Figure 4-11 Reverse I-V characteristics of rectifiers with oval shaped contacts (0.04
mm2), as a function of extent of metal overlap.


















A
. I

r







metal overlap: 2 gm
Metal overlap: 4 gm
metal overlap: 6 gm


10-11 k
I l I l l


SiC Schottky Diodes
(Oval)
area:0.04 mm
JTE:40 pm


I I I


-900 -800 -700 -600 -500 -400 -300 -200 -100 0
Bias Voltage (V)


Figure 4-12 Reverse I-V characteristics of rectifiers with oval-shaped contacts (0.04
mm2), as a function of extent of metal overlap.


10-10


I .


I







81





820 i i

800 SiC Schottky Diodes
area:0.04 mm2
780 A

S760

a 740

5 720 -- JTE: 60 pim
S- -JTE: 40 pm
I 700
0 o A--JTE: 20 pm
S 680

m 660 -

640

620

2 3 4 5 6
Metal Overlap (pm)


Figure 4-13 Variation of VB with extent of metal overlap for different JTE lengths.









contact of e-beam deposited Ni annealed at 970 OC for 3 min (contact resistivity of 1.2 x

10-6 Qcm2), was used as the ohmic contact to the substrate. SiO2 layers 3000 A were

deposited by plasma enhanced chemical vapor deposition (PECVD) and were employed

as part of dielectric overlap edge termination. Mesas were formed by dry etching. Holes

were also opened in the SiO2 stack by a combination of wet and dry etching and Schottky

contacts of e-beam evaporated (1000 A thick) Ni patterned by lift-off. The contact area

was varied from 0.01-0.36 mm2, while the mesa length was varied from 0-60 |tm and the

extent of metal overlap from 0-20 |tm. All current-voltage (I-V) measurements were

carried out at 25 C in air.

4.3.3 Results and Discussion

Figure 4-14 shows reverse I-V characteristics as a function of rectifier active area

for a fixed mesa length of 20 |tm and a metal overlap distance of 4 itm. The reverse

breakdown voltage shows a trend of decreasing as the area increases. This is most likely a

result of the greater probability for having defects such as micropipes within the active

region of the rectifier as the area increases. Therefore, our devices still have breakdown

characteristics dominated by defect density rather than bulk breakdown mechanisms. For

the smallest rectifiers the best on-state resistance (RON) obtained was 20 mQcm2,

leading to a maximum power figure-of-merit VB2/RON of 84.5 MW.cm-2 This is roughly a

factor of 6 higher than we reported previously for 4H-SiC p-i-n rectifiers fabricated on

thinner (- 12 [tm) n-layers [136].

The dependence of VB on diode area is shown in Figure 4-15 for fixed mesa length

of 20 |tm and metal overlap distance of 4 |tm. The fall-off in VB with diode area is most

pronounced at small areas (0.01-0.04 mm2) and most likely defines the region where









there is a probability of having either zero or one "killer-defect" within the device.

Compared to the theoretical best performance for 4H-SiC rectifiers of this structure, there

is still considerable improvement possible in both VB and RON [84].

The reverse I-V characteristics as a function of metal overlap distance for fixed

mesa extension length of 0 |tm and fixed diode area of 0.04 mm2 are shown in Figure 4-

16. VB goes through a maximum with increasing metal overlap distance, which is a

somewhat different trend than reported previously where it continuously decreased as the

extension distance increased [136]. However, a difference in that case was the mesa

length was fixed at 60 |tm [136]. In the present case, the trend may result from onset of

breakdown in the oxide or extension of the depletion region to include defects that initiate

breakdown.

Figure 4-17 shows the variation of VB with metal overlap distance for the same

diodes whose I-V characteristics are shown in Figure 4-16. It is clear that this parameter

has a strong influence on the breakdown voltage, with VB ranging from a low of -800 V

to a maximum of -1200 V depending on metal overlap distance.

The reverse I-V characteristics from rectifiers with fixed area and extent of metal

overlap are shown in Figure 4-18 as a function of mesa extension length. Over a wide

range of voltage, the reverse current density was found to vary as the square root of the

reverse voltage, which is characteristic of space-charge generation currents. The relation

between generation current Jo and effective generation lifetime (TG) is given by:


JG =e( 2e ) 5(n' )V05
eNB TG









where e is the electronic charge, ; the permittivity of 4H-SiC, NB the drift layer doping

concentration, ni the extrinsic carrier concentration at the measurement temperature and

V is the applied reverse bias [87]. Given a typical value of G =100 ns [87], this relation

yields a calculated generation current much lower than the measured value, suggesting

that other mechanisms such as surface leakage is playing a role. The larger area devices

should lower reverse current densities, indicating that leakage around the contact

periphery is the main contributor, as reported previously [87]. Figure 4-19 shows the

variation of VB with mesa length, extracted from the data of Figure 4-18. There is no

clear dependence of reverse breakdown on this parameter under our conditions.

Finally, Figure 4-20 shows a typical forward I-V characteristic from a small area

rectifier. The I-V characteristic is well represented by the relation:

SeV eV
I = I nkT + e n2kT

where IR and ID are the saturation currents for recombination and diffusion current

mechanisms, respectively, e is the electronic charge, k is the Boltzmann's constant, V the

applied voltage, T the absolute temperature and nl and n2 are the ideality factors for

recombination and diffusion currents, respectively. At low currents, n 2, indicating that

the dominant transport mechanism is recombination. At higher currents, n 1-1.5,

indicating that the dominant mechanism is diffusion. A typical on/off current ratio was ~

104 at +3.5V/-1000V.

4.3.4 Summary and Conclusions

In conclusion mesa edge termination of p-n 4H-SiC rectifiers passivated with SiO2

is simple alternative to use of implanted junction termination extension (JTE) methods.

The mesa extension length did not have a significant impact on the reverse breakdown









voltage of our devices, whereas increasing metal overlap distance tended to decrease VB.

On/off current ratios >105 were achieved, with VF values -25% higher than the

theoretical minimum.

In another study, 4H-SiC rectifiers were fabricated with epi JTE regions and

dielectric overlap for edge termination. Contact areas from 0.04-0.64 mm2 were

examined, with the reverse breakdown voltage being inversely dependent on contact area.

Oval and circular diodes showed higher VB than square contacts of the same area. The

metal overlap extent did not have a strong influence on VB for JTE lengths up to 40 jrm.

4H-SiC p-i-n rectifiers with VB~ -1000 V were also demonstrated. The role of

device area, mesa length and metal overlap distance on VB were examined. The

breakdown voltage was inversely dependent on device area but was not a function of

mesa length. On/off current ratios of- 104 were achieved at 3.5/-1000V. VB was a

maximum at 5 |tm metal overlap distance.












A


:- ^
-






--- area:
- -area:
S-A area:
Area:


10-5




10-7




10-9




10-11


SiC p-i-n diodes
Mesa length: 20 Lpm
Metal overlap: 4 pm







``"U


-200


Figure 4-14 Reverse I-V characteristics of p-i-n rectifiers as a function of active area.


0.36 mmr
0.16 mm2
0.04 mm2
0.01 mm2


-1000 -800 -600 -400
Bias Voltage (V)















1400

1200

1000

800

S600

400 SiC p-i-n diodes
Mesa length: 20 im
200 Metal Overlap: 4 gm


0.0 0.1 0.2 0.3 0.4

Diode Area (mm2)


Figure 4-15 VB as a function of diode area for p-i-n rectifiers with mesa length 20 ptm and
metal overlap 4 ptm.