«Copyright by Uttiya Chowdhury The Dissertation Committee for Uttiya Chowdhury certiﬁes that this is the approved version of the following ...»
The Dissertation Committee for Uttiya Chowdhury
certiﬁes that this is the approved version of the following dissertation:
MOCVD Growth for UV Photodetectors and
Light Emitting Diodes
Russell D. Dupuis, Supervisor
Joe C. Campbell
Nathan F. Gardner
John B. Goodenough
Leonard F. Register
MOCVD Growth for UV Photodetectors and
Light Emitting Diodes
Uttiya Chowdhury, B.Sc.Engg., M.S.
Dissertation Presented to the Faculty of the Graduate School of The University of Texas at Austin in Partial Fulﬁllment of the Requirements for the Degree of Doctor of Philosophy The University of Texas at Austin December 2002 To my parents: Urna and Tapan Acknowledgments I would like to express my gratitude to those who have helped make this work possible. First, I thank my advisor, Professor Russell Dupuis, for his guidance and for giving me the opportunity to participate in research in his laboratory. I would also like to thank Professor Joe Campbell and Professor Leonard Register for useful discussions throughout the years.
I have had the privilege of the guidance from some very knowledgeable researchers — I would like to thank Dr. Chris Eiting, Dr. Paul Grudowski, Dr. Adrian Holmes, Dr. Ki-Soo Kim, Dr. Ho-Ki Kwon, Dr. Damien Lambert, Dr. Joongseo Park, Dr. Jae-Hyun Ryou, Yuichi Sasajima, Dr. Bryan Shelton, Dr. Delphine Sicault, and Dr. Xuebing Zhang for the theoretical and operational knowledge they have shared with me. I would also like to mention the remaining members of the MOCVD research group with whom I have had the pleasure of working: Jin-Ho Choi, Jonathan Denyszyn, Mathilde Gobet, Richard Heller, Min-Soo Noh, Michael Wong, Dongwon Yoo, and Ting Gang Zhu. I especially appreciate the diligent assistance and sincere cooperation of the current members of the nitride team: Dr. Sicault, Michael, Ting v Gang, Jonathan, Jin-Ho, and Dongwon — without them, I could not have accomplished this dissertation.
I thank the staﬀ at the Microelectronics and Engineering Research Building — Joyce Kokes, Donna Larson and Steve Moore in the purchasing and receiving department, Jean Toll in administrative department, and William Fordyce, James Hitzfelder, Jesse James, Terry Mattord, Elton Sakewitz, Steve Schaper and Robert Stephens in the maintenance department — for providing me with a seamless work environment while they handled all the ancillary activities. I also thank Dana Dupuis for her assistance with preparing this dissertation as well as all of my other publications.
Many collaborators have contributed to the body of the work presented here. Two research teams have contributed signiﬁcantly to the success of the UV photodetector work presented here. Professor Joe Campbell and his students — Ariane Beck, Dr. Charles Collins, Xiangyi Guo, and Bo Yang have processed the epitaxial structures and given rapid feedback. The second team, the researchers at BAE Systems — Dr. Chris Cooke, Dr. Allen Hairston, Dr. Phil Lamarre, Dr. Marion Reine, and Dr. Kwok Wong have fabricated photodetector arrays of record performance besides also processing and testing discrete devices.
Dr. Ling Zhou of Lumileds Lighting, LLC for generous help with nitride light emitter work — by performing miscellaneous characterizations and by providing useful advice through hours of discussion they made a signiﬁcant
Finally, I would like to thank my parents, Urna Chowdhury and Dr.
Tapan K. Chowdhury and my sister Keya Chowdhury for their endless support and encouragement.
Due to a number of commercial, scientiﬁc and defense applications, there exists a high demand for solid-state ultraviolet (UV) light emitters and photodetectors. Group III-nitrides are very promising materials for fabrication of these because of their large direct bandgap and high thermal stability.
MOCVD growth of III-nitrides along with characterization of the material has been performed aimed towards development of UV optoelectronic devices. Photodetector arrays operating around 280 nm wavelength range have been fabricated from epitaxial structures grown and UV LEDs with wavelength shorter than 300 nm have been fabricated.
Technological challenges in improvement of these devices have been identiﬁed and roadmap for future research has been proposed.
The separation of quasi-Fermi levels in a forward biased p − n 2.2 junction. In the n-side diﬀusion region, the hole concentration
4.3 TEM image of thick AlGaN layer in an SBD device showing dislocations and reduction of dislocation density with epitaxial growth (sample M2510) (Measurement by D. N. Zakharov and
4.10 EQE spectra of fabricated SBD devices at diﬀerent bias voltages. Inset shows corresponding responsivity spectra at zero volt bias. (Measurement by P. Lamarre, M. Reine, and K.
4.11 Responsivity plot for a 256 × 256 SBD array. The color bar on the right hand side indicates responsivity in an arbitrary linear scale. (Measurement by A. Hairston, and C. Cooke; BAE
Introduction to Group-III Nitrides
1.1 Introduction Semiconductor research at the present time is mostly concentrated in the study of silicon and related materials and III-V semiconductors (arsenides, phosphides, antimonides and nitrides of gallium, indium and aluminum). It has been true for quite some time that by shear volume, silicon technology has dominated the semiconductor device marketplace. Partly due to the performance of this technology and partly due to the knowledge and technology established already by research worth signiﬁcant amount of time and money, alternate semiconductors of greater promise have not been able to enter the marketplace. The main industrial interest in other semiconductors is thus in ﬁelds where silicon technology cannot deliver — e.g. in the ﬁeld of optoelectronics and electronic devices requiring speed, power or thermal stability beyond the capability of Si. Compound semiconductor research, particularly the research in the III-V semiconductors, is concentrated in addressing device markets currently inaccessible to Si technology. Direct-bandgap III-V semiconductors span the entire visible spectral range. In addition, intrinsic high carrier mobility and the ability to form a heterostructure give them a very big advantage in fabrication of high-speed/high-power electronic devices.
It goes without saying that of the human senses, vision is the most powerful and versatile. The ability to cater to the sense of vision is clearly the basis for the greatest number of applications of optoelectronics. Directbandgap semiconductors can easily produce photons by the recombination of an electron in the conduction band with a hole in the valence band. Compared to other sources of light, semiconductor emitters are much more convenient and attractive due to their spectral purity, speed, compactness, high eﬃciency, low-power consumption, low-voltage operation, etc. For the purpose of visible illumination, semiconductor optoelectronics can be used for indicator lights, color display panels, general-purpose illumination, photodetection, etc. Besides the applications in the visible spectra, due to high speed, ease of maneuverability and a variety of other reasons, industry has an endless list of applications for virtually all wavelength ranges of the photon spectra.
The III-V semiconductors have traditionally been divided into two parts — the conventional III-Vs are the arsenides, phosphides and antimonides while the nitrides are treated separately due to diﬀerences in growth conditions, crystal structure and miscibility. The present status of the conventional IIIV semiconductor technology is clearly far ahead of the III-nitrides largely because of the unavailability of substrates for homoepitaxy and diﬃculty in growth chemistry for the III-nitrides.
The conventional III-Vs have a relatively smaller bandgap than the nitrides. This small bandgap is associated, on one hand, with a high carrier mobility, and on the other hand, with a comparative chemical and structural unstability. These semiconductors are easily broken, chemically altered and etched. The small bandgap lend them useful for light-emitting applications from green to infrared and in high-speed electronic switching applications.
The visible spectra on the blue side beyond green and the ultra-violet spectra are only addressed by the III-nitrides due to their large bandgap properties.
Also, the III-nitrides are chemically and structurally much more stable making them suitable for high temperature and corrosive environments. Like most wide bandgap semiconductors, the nitrides exhibit superior radiation hardness compared to the other smaller bandgap counterparts such as GaAs and Si — allowing their use in demanding space applications.
Although the bandgap of the nitrides extends well into the red of the visible spectra, nitride research has traditionally not paid much attention to the part of the spectra from green to red due to the relative diﬃculty in nitride technology and ready availability of alternative conventional III-V technology.
1.2 Application of UV Optoelectronics Electromagnetic radiation in the wavelength range of ∼ 400 nm to ∼ 700 nm in vacuum forms the visible range of the photon spectrum. Radiation of shorter wavelength, situated beyond the visible spectrum and spanning to the X-ray spectrum (roughly from 100 nm to 400 nm), forms the ultraviolet (UV) spectrum of light. Just like any other wavelength range of the optical spectra, light of the UV range ﬁnds a wide variety of applications in modern science and technology. While some of these applications such as curing of paint, developing of printing plates, germ decontamination, etc. are best addressed by high-power UV lamps such as mercury arc lamps, semiconductor UV optoelectronics are the ideal solution in many other areas due to smaller size, higher eﬃciency, longer life, and higher spectral purity.
UV light sources are attractive in applications requiring a small beam spot, such as laser printing and optical data storage, because of its shorter wavelength compared to visible light. Combined with suitable ﬂuorescent material, a UV light source can be used to generate white light (of broad spectral range in the visible region) or light of a certain color (having a narrow spectral range). UV light source and detectors can be used in a variety of military or defense related applications including detection of biological agents from their ﬂuorescence signature and detection of missiles from the UV emission of the exhaust plume.
For defense related purposes, one very important spectral range is the solar-blind range in the 240 – 280 nm. Solar radiation energy of wavelength shorter than ∼ 290 nm is strongly absorbed by the terrestrial ozone layer and at Earth’s surface, a photodetection system operating in this spectral range will not see a huge background signal from the solar radiation. A photodetector array system sensitive to this band situated in the space beyond the ozone layer on the dark side of earth will be an extremely sensitive detection system for objects such as missiles, planes and rockets coming out of the environment.
Also, line-of-sight space-to-space communication occurring beyond the ozone layer in this band would not be detectible from the Earth. Ground based lineof-sight communication systems based on UV lasers and detectors are also attractive for various military and commercial applications.
Apart from these, there are many commercial, scientiﬁc and military applications of compact solid-state UV light sources and high-eﬃciency UV photodetectors. As the technology matures, more applications and demand for UV optoelectronics are sure to be developed.
1.3 Properties of Nitride Semiconductors Group-III nitrides (InN, GaN, AlN and their alloys) are semiconductors crystallizing in the wurtzite structure (space group 186: P 63 mc) in their most stable form. In this form, all the binary and alloy III-nitrides are directbandgap materials with room-temperature bandgaps ranging from ∼ 1.89 eV for InN (corresponding to a vacuum wavelength ∼ 656 nm red luminescence) to ∼ 6.2 eV for AlN (vacuum wavelength ∼ 200 nm, deep ultraviolet). The ratio of c to a-lattice constant varies for the system and is not equal to the ideal wurtzite ratio of 1.633 for any of the nitride alloys. The III-nitrides also crystallize in the zinc-blende structure — the zinc-blende energy gaps being slightly lower. Also, unlike in the wurtzite form, the zinc-blende form of AlN does not have a direct bandgap. The scope of the present work is solely conﬁned to the study of nitrides in the wurtzite structure.
Due to the strong ionic component of the bond between the Group-III atom and the nitrogen atom, the III-nitrides are highly stable structurally and chemically. GaN and AlN display superior resistance to most chemicals and of conventional wet etchants are known to etch in hot KOH solution only. The chemical inertness can, on one hand, be advantageous because the device will be able to tolerate chemically harsh environments. However, this also makes processing steps much harder. Particularly, one very useful aspect of conventional III-V technology is the ability to selectively etch down to atomically smooth layers.
Electrical properties of III-nitrides have also been extensively studied.
Unintentionally doped GaN is usually found to be n-type in the low 1016 /cm3 concentration range although optimized growth conditions can yield semiinsulating material. The most common n-type dopant for GaN is Si. The only successful p-type dopant remains Mg, which is a very deep acceptor with an acceptor energy in the range of 200 meV in GaN. This leads to a roomtemperature ionization of only about 1% of the acceptor atoms, requiring a Mg concentration in the 1020 /cm3 for a free hole concentration of 1018 /cm3.