UV Photodetectors, Focal Plane Arrays, and Avalanche Photodiodes
A DISSERTATION
SUBMITTED TO THE GRADUATE SCHOOL
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS for the degree
DOCTOR OF PHILOSOPHY
Field of Electrical and Computer Engineering
By
Ryan M
cClintock
EVANSTON, ILLINOIS
June 2007
2
© Copyright by Ryan McClintock 2007 All Rights Reserved
3
Abstract
UV Photodetectors, Focal Plane Arrays, and Avalanche Photodiodes
Ryan M
cClintock
The study of III-Nitride based optoelectronics devices is a maturing field, but there are still many underdeveloped areas in which to make a contribution of new and original research.
This work specifically targets the goals of realizing high-efficiency back-illuminated solar- blind photodetectors, solar-blind focal plane arrays, and visible- and solar-blind Avalanche photodiodes. Achieving these goals has required systematic development of the material growth and characterization, device modeling and design, device fabrication and processing, and the device testing and qualification. This work describes the research conducted and presents relevant devices results.
The AlGaN material system has a tunable direct bandgap that is ideally suited to detection of ultraviolet light, however this material system suffers from several key issues, making realization of high-efficiency photodetectors difficult: large dislocation densities, low n-type and p-type doping efficiency, and lattice and thermal expansion mismatches leading to cracking of the material. All of these problems are exacerbated by the increased aluminum
4
compositions necessary in back-illuminated and solar-blind devices. Overcoming these obstacles has required extensive development and optimization of the material growth techniques necessary: this includes everything from the growth of the buffer and template, to the growth of the active region.
The broad area devices realized in this work demonstrate a quantum efficiency that is among the highest ever reported for a back-illuminated solar-blind photodetector (responsivity of 157 mA/W at 280nm, external quantum efficiency of 68%). Taking advantage of the back illuminated nature of these detectors, we have successfully developed the technology to hybridize and test a solar-blind focal plane array camera. The initial focal plane array shows good uniformity and reasonable operability, and several images from this first camera are presented. However, in order to improve the performance of these devices to the point where they can effectively compete with photo-multiplier tube technology, it is necessary to develop devices with internal gain. To this end GaN and AlGaN based avalanche photodiodes have been studied, and we report the first realization of a solar-blind back-illuminated avalanche photodiode. The next logical step is to continue this work and realize Geiger mode avalanche photodiodes capable of single photon detection.
5
Acknowledgements
I would like to first thank my research advisor Professor Manijeh Razeghi for her continued support and guidance throughout both my undergraduate and graduate education here at Northwestern University. Without her, and the wonderful research facilities she has provided here at the Center for Quantum Devices, this work would never have been possible.
I would also like to thank my committee members for agreeing to participate in my defense, and would like to acknowledge their support and encouragement.
I would like to acknowledge the federal government for providing three years of support under a National Defense Science and Education Graduate fellowship. As well, I would like to acknowledge Northwestern University for supporting my fourth year with a University
Fellowship. And I would finally like to acknowledge the Richter Trust for supporting me during my dissertation year with a Richter Trust Fellowship.
I would also like to specially thank my close colleagues studying III-Nitrides: Kathryn Minder, Dr. Patrick Kung, Dr. Jose L. Pau, Alireza Yasan, and Can Bayram Without their assistance and support, much of this work would not have been possible. I would also like to acknowledge all of the other graduate students, both past and present, at the Center for Quantum Devices.
Finally I would like to thank my parents: Dr. Marty McClintock and Mrs. Nancy McClintock. To them I am eternally grateful, and I doubt that I will ever fully understand everything they have done for me.
6
i. Table of Contents
Abstract ...3
Acknowledgements ...5
i. Table of Contents...6
ii. List of Figures ...9
iii. List of Tables...19
1. Introduction ...20
2. Background ...23
2.1. The Solar Ultraviolet Spectrum 23 2.2. UV Photodetector Applications 25 2.3. UV Detection Technologies 31 2.4. Historic Development of III-Nitride Based UV Photodetectors 34 2.4.1. Photoconductors 35 2.4.2. Schottky Metal-Semiconductor-Metal Detectors 37 2.4.3. Schottky Barrier Photodiodes 40 2.4.4. Photocathodes 41 2.4.5. Front Illuminated p-i-n Photodiodes 42 2.4.6. Back Illuminated p-i-n Photodiodes 45 2.4.7. Avalanche photodiodes 48 3. Important UV Photodetector Characteristics...52
3.1. General Photodetector Parameters 52 3.2. Basic Noise Analysis Theory 54 3.3. Noise Analysis in AlGaN p-i-n Photodiodes 59 3.4. Avalanche Photodiode Parameters 61 3.5. Noise Analysis in Avalanche Photodiodes 65 4. Wide Band-Gap III-Nitride Material Growth...69
4.1. Introduction 69 4.2. Metal Organic Chemical Vapor Deposition, an Overview 69 4.3. Growth Nucleation: Low and Intermediate Temperature Buffers 78 4.4. Atomic Layer Epitaxy for Growth of AlN and AlGaN 81 4.4.1. Growth of AlN by ALE 82 4.4.2. Growth of AlGaN / AlN Superlattice Template 85 4.5. Growth and Doping of Wide-Bandgap AlGaN Material 89 4.5.1. N-type Doping of AlGaN 89 4.5.2. P-Type Doping of AlGaN 92 5. Experimental Procedure: Large Area Single Element Detectors ...95
5.1. Introduction 95 5.2. Material Growth and Characterization 96
5.3. p-i-n Photodetector Processing Overview 104
7
5.4. Photodetector Measurement and Discussion 105 5.5. Deep UV Back-Illuminated Photodetectors (255 nm) 111
6. Experimental Procedure: Solar-Blind Focal Plane Arrays ...114
6.1. Introduction 114
6.1.1. Focal Plane Array Technology 115 6.1.2. Historic Development of UV Focal Plane Arrays 117
6.2. Material Growth and Characterization 119
6.3. FPA pixel characteristics 120 6.4. FPA Processing Overview 124 6.5. FPA Images and Discussion 126 6.6. Improvement of Device Performance 129
7. Experimental Procedure: AlGaN Based Avalanche Photodiodes...133
7.1. Introduction 133 7.2. Material growth and device processing 135
7.2.1. Material Growth 135
7.2.2. Device Processing 137
7.3. Unbiased device performance 138
7.3.1. Current-voltage characteristics at low bias 138
7.3.2. Photoresponse 141
7.4. Avalanche Mode device operation 142
7.4.1. Current-voltage curves under bias 142
7.4.2. Device Modeling 146
7.4.3. Ruling out other origins for the gain 149
7.5. Conclusion 149 8. Experimental Procedure: GaN Based Avalanche Photodiodes...151
8.1. Introduction 151 8.2. Material Growth 152 8.3. Device Processing 155 8.4. Device Results 156
8.4.1. I-V measurements 156
8.4.2. Gain Measurements 162
8.4.3. Origins of the observed multiplication 167
8.4.4. Noise Analysis 168
8.5. Conclusion 170 9. Future Work: Geiger Mode APDs...172
9.1. Introduction 172 9.2. Preliminary Device Results 173
10. Appendices ...177
10.1. Appendix 1: Material Characterization Techniques 177
10.1.1. Structural Characterization 177 10.1.2. Optical Characterization 189 10.1.3. Electrical Characterization 195
10.2. Appendix 2: Device Testing and Characterization 200
10.2.1. Current Voltage Measurement 200 10.2.2. Responsivity and External Quantum Efficiency 203
8 10.2.3. Bias Dependant Responsivity 209
10.2.4. Gain Measurement 210
10.2.5. Noise Measurement 211
10.2.6. Geiger Mode APD Measurement 213
10.3. Appendix 3: ROIC Specifications and Camera Interfacing 216
10.3.1. ROIC Specifications 216 10.3.2. Camera System Electronics 220
10.3.3. Imaging Optics 223
10.3.4. Imaging Software and Image Correction 227
10.4. Appendix 4: Sample UV FPA movies 231
10.4.1. Movie 1: Moving CQD Logo 231 10.4.2. Movie 2: Dancing Exposed Electric Arc 232 10.4.3. Movie 3: Reflection from a Patterned Mirror 233
10.5. Appendix 5: Development of a Portable Camera System 234
10.5.1. Project Goal: 234
10.5.2. Software Design: 240
10.5.3. Hardware Design: 264
10.6. Appendix 6: Development of a Semiconductor Wafer Cleaning System 271
10.6.1. Systems Overview 273
10.6.2. Control and Software Design 279
11. References ...283 12. Curriculum Vita ...301
12.1. Education 301 12.2. Honors 301 12.3. Graduate Coursework 301
12.4. List of Publications 302
9
ii. List of Figures
Figure 1. Solar Spectral Irradiance as seen from space (light blue), and from sea level (orange). On earth absorption by atmospheric ozone strongly attenuates wavelengths less than ~290nm thus creating a solar blind window (data taken
from ref. 1). ...24 Figure 2. III-Nitride UV photodetectors have a number of civilian and military
applications as illustrated above...26 Figure 3. Illustrations of the operation of secure UV-based NLOS communication system
in a variety of environments...27 Figure 4. UV Absorption and fluorescence of three common biological markers (tyrosine,
tryptophan, NADH) illustrating the detector and emitter requirements. ...29 Figure 5. The quaternary AlInGaN material system is indicated in blue. For comparison
other viable semiconductor materials are shown in black. ...32 Figure 6. Detective Quantum Efficiencies of materials for potential use in UV imaging
applications. In the solar blind region of the spectrum AlGaN has a clear
advantage over both photocathodes such as Cs2Te , and UV enhanced CCDs...33 Figure 7. Alternatives to AlGaN for solar blind imaging, and their disadvantages ...34 Figure 8. Photoconductors exhibitinbg sharp cut-off wavelengths covering the entire
AlxGa1-xN compositional range: from GaN (365nm) to AlN (200 nm). The inset
shows a simplified photoconductor structure...36 Figure 9. Shows the geometry of a typical interdigitated finger MSM device with a
length of 150 μm a finger width of 2 μm and a pitch of 10 μm...38 Figure 10. Shows a typical spectral response of a Schottky based MSM photodetector. ...39 Figure 11. Shows the typical spectral response of a Schottky barrier photodetector. The
corresponding device structure is shown in the inset...41 Figure 12. Shows the typical structure of a front illuminated AlxGa1-xN based p-i-n
photodiode...43 Figure 13. The spectral response of a set of front illuminated AlxGa1-xN p-i-n
photodiodes with various Al compositions. The line at the top indicates the theoretical maximum for a p-i-n photodetector corresponding to a quantum
efficiency of one...45
10
Figure 14. The basic structure of a back illuminated p-i-n photodetector...46
Figure 15. Typical photoresponse of a back-illuminated solar blind photodetector showing the difference between the front- and back-illuminated responses. ...47
Figure 16. Mosaic image of micro-plasma luminescence from an un-optimized array of 16 diodes showing the luminescent nature of the process, and the distribution of defects. ...49
Figure 17. GaN based APD showing an avalanche multiplication of nearly 1500 at a breakdown field strength of 2.7 MV/cm. ...50
Figure 18. AlGaN-Based Solar-Blind APD showing a maximum gain of 2000 at an electric field of 7 MV/cm...51
Figure 19. The dark current of GaN p-i-n photodetector at different applied biases. ...56
Figure 20. The noise power spectral density of a GaN p-i-n photodetector...58
Figure 21. Signal, noise, and dark current contributions to the output of an APD. ...63
Figure 22. Noise power spectral density versus bias around breakdown for a GaN APD...66
Figure 23. Aixtron AIX 200/4 MOCVD reactor used for growth of III-Nitride material discussed in this work. ...70
Figure 24. Schematic diagram of the MOCVD reactor used to grow the material discussed in this work. This diagram shows the full gas handling system, the reactor growth chamber, and the vacuum system. ...72
Figure 25. Schematic diagram of the 4 step MOCVD growth process. ...74
Figure 26. Growth efficiency of Al0.4Ga0.6N at different growth pressures. The dashed lines mark the growth efficiency of 1000 mm/mole corresponding to the growth pressure of ~35 mbar...76
Figure 27. Photoluminescence for a series of AlGaN:Si layers grown at different growth pressures showing the material quality tradeoff associated with growth at lower pressures...77
Figure 28. Crystallography of AlN grown on top of sapphire. The image at the left shows a plan view. This illustrates to the 30 degree rotationof the crystal structure that leads to a 13.2 % effective lattice mismatch. The image at the right shows a sectional view showing the formation of a misfit dislocation every 8 atoms. ...79
Figure 29. High resolution TEM image of the sapphire/ AlN interface showing the generation of misfit dislocations. ...81
11
Figure 30. Diagram of the valve switching used to grow AlN by atomic layer epitaxy. ...83 Figure 31. High resolution x-ray diffraction scans from an ALE grown AlN, showing an
extremely narrow systemic (002) and narrow asymetric(105) width owing to the
high-quality of the material...84 Figure 32. AFM Image of the surface of ALE grown AlN on a low temperature AlN
buffer...84 Figure 33. Diagram of the valve switching sequence used to grow high aluminum
composition AlGaN by atomic layer epitaxial...86 Figure 34. AFM image showing the high-quality of the surface of an AlGaN/AlN SL
grown on AlN, all grown using the ALE technique...87 Figure 35. Macroscopic hexagonal etch pits are formed by the worst of the defects after
etching in a soultion of hot KOH or H3PO4. ...88 Figure 36. Activation energy of Mg in AlxGa1-xN as a function of Al mole fraction...93 Figure 37. AFM image showing the surface of the high-quality AlGaN/AlN SL template,
3 nm data scale. The RMS roughness for the 5 mm square scan shown above is
1.3 Å...97 Figure 38. Optical transmission and absorbance squared of the Al0.5Ga0.5N:Si:In
conduction layer grown on a high-quality AlN/AlGaN SL template. The conduction layer shows a sharp cut off owning to the excellent material quality,
with the absorption edge occurring at 260 nm. ...100 Figure 39. Schematic cross-section showing the structure of a back-illuminated solar-
blind photodetector. ...101 Figure 40. Responsivity vs. wavelength for a Back illuminated p-i-n photodiode,
showing a significant negative photoresponse...102 Figure 41. I-V curves for several different Cp2Mg flows used to optimize the p-type
layers. The structure and measurement geometry are shown in the lower right
hand corner...103 Figure 42. Optical micrograph of a 1mm x 1mm Solar-blind photodetectors showing the
device layout, and the proximity of the n-contact...104 Figure 43. I-V curve of a typical back illuminated photodiode in semi-log scale. ...106 Figure 44. Natural log of current versus voltage. The low current regime is modeled and
an ideality factor of 2.89 is extracted...107
12
Figure 45. Responsivity vs. wavelength for a typical photodiode, showing a peak responsivity of 150 mA/W at a wavelength of 280 nm, and no negative photoresponse. This peak responsivity corresponds to an external quantum
efficiency of 68%. ...108
Figure 46. Shows the external quantum efficiency of the device in linear scale. ...109
Figure 47. Schematic diagram of the reflective and transmission losses in our device. ...110
Figure 48. 255 nm deep UV photodetector device structure...111
Figure 49. Unbiased responsivity in log scale (left) and the corresponding unbiased external quantum efficiency in linear scale (right). ...113
Figure 50. Shows a schematic cross-section of a III-Nitride FPA flip-chip bonded to a Si ROIC. ...116
Figure 51. Shows a simplified schematic of a typical CITA unit cell as used in a ROIC...116
Figure 52. I-V curve of a single 25 μm × 25 μm FPA pixel shown in linear scale. The inset shows the same data in logarithmic scale...121
Figure 53. I-V statistics for the array, showing the good current spreading. ...122
Figure 54. Unbiased responsivity in log scale (left) and the corresponding unbiased external quantum efficiency in linear scale (right) from a representative 25 μm x 25 μm pixel. ...123
Figure 55. Electron micrograph of the focal plane array before bonding to the ROIC: after lithographic patterning, metallization, and deposition of the indium bumps. ...124
Figure 56. 320 × 256 Image of: A) a paper-cutout in front of a Germicidal lamp and B) an electric arc as seen with the solar-blind FPA camera...126
Figure 57. Image of a human likeness as seen with the solar-blind FPA camera. ...127
Figure 58. Histogram of Pixel intensity for a flat field of UV illumination. This figure shows the variation in pixel response within the array. ...128
Figure 59. Cross-sectional diagram of the current and proposed FPA designs. The top half shows the current design illustrating how a significant amount of light reaches the ROIC, the bottom half shows the proposed solution that will make the FPA 100% opaque...130 Figure 60. Top view of an FPA processed with a dark-field layer. Gold is the top contact,
and the maroon area surrounding the top contacts in the black-out layer, it
13
almost overlaps the top contact and blocks out 99% of the area that would
otherwise be exposed. ...131 Figure 61. 15 μm tall indium bump after reflow, seen from an oblique angle...132 Figure 62. Comparison of exiting UV detection technologies to AlGaN based detectors,
showing the gap between the detectivity of PMTs and AlGaN based devices,
and the need for APDs (in red). (After references and ) ...134 Figure 63. Schematic cross-section showing the device structure of the back-illuminated
APD...136 Figure 64. Scanning electron micrograph of an APD after processing. The common n-
contact(not shown) is far removed from the mesas to avoid air breakdown of the
devices...138 Figure 65. Comparison of Dark current density between 1mm x 1mm devices and 25 μm
x 25 μm devices, showing the disparity in scaling of the dark current...139 Figure 66. Left) Current-Voltage curve showing the current density in log scale. Right)
Natural log of current and fit to linear region used to extract an ideality factor
(n) of 2.8...140 Figure 67. Left) Unbiased responsivity shown in log scale. Right) External quantum
efficiency shown in linear scale. ...142 Figure 68. Current-voltage behavior as a function of applied reverse bias, both under
illumination and in the dark. Errors bars have been added to indicate the variation over 3 consecutive pairs of alternating light-dark measurements of the
same diode...143 Figure 69. Calculated photocurrent (left axis) and corresponding gain (right axis) from
the data of Figure 68. Error bars indicate +/- 1 standard deviation. Breakdown is taken at 40 volts. ...144 Figure 70. Schematic diagram of parallel and sequential ionization. Sequential ionization
(left) can lead to Geiger mode operation, however parallel (right) restricts the
multiplication to a geometric increase. ...146 Figure 71. Electric Field Profile under various applied reverse biases. ...147 Figure 72. Avalanche gain model of the device. The solid curve shows the experimental
data, the dashed curve shows the model. ...148 Figure 73. Atomic force microscopy (AFM) imaging of the surface of a high-quality
GaN layer grown on an AlN template...153
14
Figure 74. High-resolution x-ray diffraction of GaN on an AlN template layer...154 Figure 75. Schematic diagram of GaN APDs on AlN templates (left) and GaN templates
(right). ...155 Figure 76. Schematic diagram of the APD as processed...156 Figure 77. Breakdown voltage as a function of intrinsic layer thickness used to extract a
critical electric field strength of 2.73 MV/cm...158 Figure 78. Model of the electric field profile across the device as a function of the applied
reverse bias: experimentally breakdown occurred at 102 V corresponding to 3.2
MV/cm in the model. ...159 Figure 79. Breakdown characteristics of samples A, B, and C are shown. Inset: The
experimental breakdown voltages obtained for different thicknesses of the
intrinsic layer...161 Figure 80. Variation of dark current of a GaN APD biased 2V above breakdown. The
standard deviation is less than 6% over more than 60 hours. ...162 Figure 81. Multiplication factors for electrons (Mn) and holes (Mp) obtained from sample
A (left) and sample B (right)...164 Figure 82. Solid lines: ionization factors obtained for electrons (αn) and holes (βp) from
experiment. The dashed lines represent theoretical values for βp and αn, as
extracted from ref 180...165 Figure 83. Evolution of the spectral response of a GaN APD near breakdown. Spectra are
shown on the left, and the evolution at three selected wavelengths are shown on
the right. ...166 Figure 84. Evolution of the device breakdown voltage with temperature. ...167 Figure 85. The Spectral Power Density (Sn) of sample A at the onset of breakdown is
shown from 91 V to 102 V with 1 V steps. The two narrow spikes at 60 and 120 Hz correspond to line noise...169 Figure 86. Left) Spectral power density is plotted as a function of total current for sample
A under front- (triangles) and back-illumination (squares). Right) calculated
excess noise factors for front- and back-illumination...170 Figure 87. (Top) Schematic diagram of the passive quenching circuit used to apply the
DC bias and AC excitation pulse to the Geiger mode APDs...174 Figure 88. Geiger-mode spectral response of a GaN APD detecting at the 10 photons-per-
pulse level. ...175
15
Figure 89. Left) Linear mode photoresponse at 75V reverse bias. Right) Geiger mode
photoresponse with 10 Photons/pulse illumination. ...176 Figure 90. A.) Optical Microscope used to investigate broad area surface morphology B.)
Representative optical micrograph taken at a magnification of 100x, showing a
50% AlGaN sample with a large number of cracks...179 Figure 91. Custom Designed software created for capturing of still images using an
optical microscope. ...180 Figure 92. Schematic diagram of a scanning electron microscope showing the electron
source, the lenses and scanning coils, the sample under test, and the electron
detector (based on ref. )...181 Figure 93. Schematic diagram of an atomic force microscope (Based on Digital
Instrument Multimode SPM manuals)...184 Figure 94. Diffraction from a set of crystal planes according to Bragg’s law. ...187 Figure 95. UV photoluminescence setup showing the three laser excitation sources that
can be used to stimulate PL, as well as the sample stage, focusing optics and
monochromator. ...190 Figure 96. Custom Labview software developed as part of this work to facilitate the
measurement and analysis of routine Photoluminescence...192 Figure 97. Ultraviolet transmission measurement system showing the xenon lamp,
monochromator, chopper, lens, sample holder, and the UV-enhanced silicon
photodetector...193 Figure 98. Custom Labview software developed as part of this work to measure UV
transmission. ...194 Figure 99. Custom Hall effect measurement electronics assembled to facilitate the
measurement of high impedance III-Nitride materials. ...196 Figure 100. Schematic diagram of a typical hall mobility measurement system showing
the van der Pauw contact geometry as used for measurement of the resitivity (A)
and for the determination of the hall coefficient (B)...198 Figure 101. Current-Voltage characterization setup used to measure and record I-V
curves. The probe station has probe-tips as small as .5 μm, and is suitable for
probing to individual pixels of an unbounded FPA. ...201 Figure 102. Typical IV curve of a single 25μm x 25μm FPA pixel shown in log scale.
The linear fit of this data gives an ideality factor of 3.7; slopes for ideality
factors of 1 and 2 are displayed for comparison purposes...202
16
Figure 103. Measurement setup used to characterize the photoresponse of both broad-
area photodetectors and individual text pixels from an FPA. ...205 Figure 104. Photograph of a wafer on the probe-station used for responsivity
measurements showing the fiber optical cables (above and below wafer) and the
two triaxial probes...206 Figure 105. Custom written software used to make UV detector responsivity and external
quantum efficiency measurements...207 Figure 106. Xenon lamp flux as a function of time since turning on the lamp showing the
variation during lamp warm up. The blue bars indicate a +/- 1% variation in
xenon lamp power...208 Figure 107. Example of permutation of data from gain measurements used to generate
gain data and error bars...211 Figure 108. Custom software written to facilitate the collection of noise Spectral Power
Density as a function of reverse bias for characterization of photodetectors and
APDs. ...213 Figure 109. Left) diagram showing the self quenching that the 100kΩ resistor ideally
provides to the APD, Right) Schematic diagram of the Geiger mode APD
biasing circuit...214 Figure 110. Indigo ROIC with 320 x 256 FPA bonded to it. The chip is only ~1 cm x
1cm in size and contains almost all of the electronics necessary to operate as a
solar-blind UV imager...217 Figure 111. General overview of ROIC specifications and a list of the imaging
conditions used to obtain all of the images presented within the context of this
work. ...218 Figure 112. Focal plane array camera system used to record images presented in this
work. Shown with side panel removed to illustrate the internal electronics necessary to operate the ROIC. NOTE: Although a Dewar is shown, the FPA is
only operated at room temperature under an ambient atmosphere. ...222 Figure 113. Transmission spectrum of the 280 nm band-pass filter used in camera head
optics. ...224 Figure 114. Schematic diagram of the optics comprising the solar-blind imaging system
used to obtain the images presented in this work...225 Figure 115. Schematic Diagram of the optics used to record reflective UV images using
the solar-blind focal plane array...226
17
Figure 116. Schematic diagram of the imaging optics used to record images for emissive
UV sources, such as an electric arc...227
Figure 117. This figure shows a single frame from the middle of the attached movie #1. ...231
Figure 118. This figure shows the first frame of the attached movie #2...232
Figure 119. This figure shows the first frame of the attached movie #3...233
Figure 120. The system is small, lightweight, and can be easily packaged for transport to diverse locations in a hard carrying case requiring only LN2 and a laptop computer for operation...234
Figure 121. Schematic diagram of the components that make up the portable camera system...236
Figure 122. Overview of the electronics package layout ...237
Figure 123. Screen capture of the various windows that make up the portable camera system user interface...238
Figure 124. Virtex 4 Mini-Module used to run the embedded portion of the software, also shows the EEPROM, RAM, and Ethernet-PHY used by the Virtex-4 FPGA...239
Figure 125. Schematic diagram showing the building blocks (and their sub components) that constitute the portable camera system software...240
Figure 126. Main program window...242
Figure 127. Right-Click pixel processing details as shown on the status bar. ...245
Figure 128. Video scope and camera setup user interface ...246
Figure 129. Background subtraction and limited bad-pixel replacement user interface...249
Figure 130. Two-point NUC and bad-pixel detection user interface ...252
Figure 131. Video capture user interface ...253
Figure 132. FPA camera system image transport packet layout. ...254
Figure 133. User interface for low-level control of programmable camera biases ...255
Figure 134. Control serial word interface for Indigo 9705 & 9809 ROICS...256
Figure 135. Windowing serial word interface for Indigo 9705 & 9809 ROICs...257
Figure 136. FPA Camera Source interface, showing diagnostic information...258
18
Figure 137. Advanced image processing user interface...259 Figure 138. Brightness and contrast user interface. Allows adjustment of gamma and
colorization...261 Figure 139. Cross-sectional view of the camera system electronics showing the front and
back planes showing the attachment to the dewar. ...265 Figure 140. Embedded CPU Daughter card and power supply board. ...266 Figure 141. Custom designed analog card with video gain pipeline, programmable offset
generator, and two channel 14 bit 20MSPS ADC. ...268 Figure 142. 6 channel programmable bias card...269 Figure 143. Clock driver and 2 channel programmable bias card...270 Figure 144. Wafer holder (center) and multi-nozzle spray arm (top) showing the
Semiconductor wafer cleaning system installed in a chemical hood...272 Figure 145. General layout of the semiconductor wafer cleaning system, showing the
location of support equipment in the chase, away from the main use interface
which occupies minimum space in the hood...273 Figure 146. Left) picture of the insulated counter-current heat exchangers where they
meet the valves. Right> schematic diagram of the three heat exchanger located
behind the hood...274 Figure 147. Left) Photo of the solvent storage tanks located within the flammable storage
cabinet. Right) Schematic diagram of the solvent pressurization plumbing
(green), flexible lines (blue) and tanks (red)...275 Figure 148. Left) Photo of bottom of wafer fixture showing water lift pump (top shelf),
and waste collection container (bottom shelf), Right) schematic diagram of the waste collection plumbing showing the lift pump draining waste water into a
nearby acids sink...278 Figure 149. Semiconductor wafer cleaning system electronics and pneumatics control
package...281 Figure 150. Main control panel for interfacing with the semiconductor wafer cleaning
system...282
19
iii. List of Tables
Table 1. Dark current and shot noise levels for various reverse bias voltage values. ...57 Table 2. Sources used in MOCVD growth and doping of Al(In)GaN...71 Table 3. Comparison of electrical properties of n-type Al0.5Ga0.5N grown with different
approaches to the doping of the material. ...90 Table 4. Table of approximate carrier concentrations of the various layers used later in
the modeling of this device structure. ...136 Table 5. Table of device structures and limited device characteristics. ...159 Table 6. Serial mode word bits and the functions they control. ...219 Table 7. List of the 14 signals necessary for the operation of an FPA based on the Indigo
ROIC ...220 Table 8. A list of the available menu functions as well as their descriptions...243
20
1. Introduction
The study of III-Nitride based optoelectronics devices is a relatively new and exciting field, and there are still many underdeveloped areas in which to make a contribution of new and original research. This work specifically targets the goals of realizing high efficiency back- illuminated solar-blind photodetectors, solar-blind focal plane arrays, and Avalanche
photodiodes. However to achieve these goals has required systematic development of the material growth and characterization, device modeling and design, device fabrication and processing, and the device testing and qualification.
This work has encompasses a time span of over 7 years, has lead to more than 26 journal articles and conference proceedings, and has even been reported in the mainstream press several times. At the start of this research GaN material growth had been well established and high efficiency top-illuminated GaN based photodetectors had already been realized.
Preliminary AlGaN growth techniques were already established at the center for quantum devices, and top-illuminated photodetectors of varying quality had already been demonstrated across the material compositional range. However, little research had really been conducted into the realization of back-illuminated photodetectors (necessary for focal-plane arrays). What back-illuminated solar-blind photodetectors did exist had vastly inferior performance, showing an eternal quantum efficiency of less than 15%. Before the start of this work we had not established any capabilities in either focal plane arrays (ultraviolet or infrared), or avalanche photodiodes.
In order to realize high quantum efficiency back-illuminated solar-blind photodetectors it was necessary to systematically develop a new approach to the growth of high-quality high-
21
aluminum composition materials. This involved complete optimization of every layer involved in the growth of back-illuminated ultraviolet photodetectors: all the way from the buffer layer at the bottom, all the way to the top contact layer. For each layer, all growth- related parameters such as the growth temperature, the growth pressure, reactor gas flows, and the V/III ratio were carefully studied and optimized to maximize the device performance. New buffer layers and novel low V/III ratio growth techniques were developed to realize a new high-quality AlN template layer growth technique that was instrumental in the realization of the high quantum-efficiency back illuminated solar-blind photodetectors, focal-plane arrays, and avalanche photodiodes. With these optimizations, and this new template, we realized devices with an external quantum efficiency of 68%; this is still among the highest values ever reported in the literature, and was a significant achievement at the time.
With the realization of high quantum efficiency back-illuminated photodetectors the work proceed to the next logical step, realization of a solar-blind focal plane array camera. This involved significant work to develop the device processing, control the device uniformity, develop of indium bump deposition and hybridization technology, and to develop testing and characterization techniques for focal-plane arrays. Due to the immense complexity and cost associated with realizing ultraviolet focal-plane array camera much of this work was carried out in conjunction with parallel work to develop infrared camera technology. However in the end this work paid off: we used a commercial read-out integrated circuit to successfully realize the first solar-blind focal plane array fabricated entirely at a university. This camera has been further extended, and as part of this work we have developed an entirely custom portable camera system to electrically operate these focal-plane arrays, and provide imaging.
22
The final stage of this work has been the development of avalanche photodiodes operating in the ultraviolet. This is an entirely original area of research that had been little explored in the past. Our material and processing improvements have allowed us to realize avalanche photodiodes. We were the first to report avalanche multiplication from a back- illuminated solar-blind avalanche photodiodes. These devices had a gain in excess of 1000, which was as high as any GaN based device reported at that time. We have gone on from here to go back and more extensively study GaN based avalanche photodiodes. We were the first to report back-illuminated p-i-n based GaN avalanche photodiodes, and are currently pursuing these for array applications. We have also made significant study of the ionization
characteristics of holes and electrons in GaN, resulting in one of the first experimental determinations of the impact ionization coefficients in GaN.
The next logical step for this research is the realization avalanche photodiodes operating in Geiger mode, and capable of single photon detection. Significant strides have already been made in this direction.
23
2. Background
2.1. The Solar Ultraviolet Spectrum
This work focuses exclusively on the ultraviolet (UV) portion of this spectrum. The majority of the solar radiation that reaches the earth lies close to the visible region of the spectrum. The intensity falls off from there slowly into the infrared and more quickly into the ultraviolet with the spectrum resembling that of a typical blackbody source with a temperature of approximately 5800 K. This solar spectral irradiance is shown in Figure 1; wherein the data is taken from reference1.
Ultraviolet light is defined as light having a wavelength less than about 400 nm, but longer than that of soft X-rays. This portion of the spectrum can be divided into four major regions: UV A covering wavelengths in the range of 400 to 315 nm, UV B covering 315 to 280, UV C, 200 to 280 nm,2 and vacuum UV or far UV, between 200 nm and ~10 nm. AlGaN based III-Nitride photodetectors are well poised to cover a large portion of the UV spectrum:
AlGaN ranges from binary GaN with a direct band gap of 3.4 eV (365 nm), to binary AlN with a band gap of 6.2 eV (200 nm). Latter in this section this versatility is demonstrated for both photoconductors and p-i-n photodiodes in Figure 8 and in Figure 13, respectively.
24
Solar Blind Window: λ= 240 to 290 nm
Figure 1. Solar Spectral Irradiance as seen from space (light blue), and from sea level (orange). On earth absorption by atmospheric ozone strongly attenuates wavelengths less than
~290nm thus creating a solar blind window (data taken from ref. 1).
In terms of UV detectors, a further distinction in made in the UV spectrum by separating devices based upon the strategic wavelength of 290 nm. Detectors with a cutoff wavelength less than 400 nm are termed visible-blind, due to their insensitivity to visible radiation. This is a desirable UV detector trait in that it reduces interference from visible light.
However as shown above in Figure 1 the solar irradiance reaching the earth is still significant from 400nm down to about 290 nm where is falls off very strongly. Below 290nm almost no light reaches the earth’s surface due to atmospheric absorption by ozone (O3) in the upper atmosphere3. This creates a universal low background window called the solar-blind window;
and because of this, UV detectors with a cutoff wavelength less than 290nm are termed solar-
25
blind detectors, and are ideal for terrestrial detection of man-made UV sources. The early III- nitride detector research focused on visible blind detectors, but recently as the growth of high Al-composition AlGaN material has matured; interest has focused more strongly on detectors operating in the solar blind region of the ultraviolet spectrum.
2.2. UV Photodetector Applications
The development of UV photodetectors has been driven by numerous applications in the defense, commercial, and scientific arenas. These include, for example, covert space-to- space communications, secure non line-of-sight communications, early missile threat detection, UV spectroscopy, chemical and biological threat detection, flame detection and monitoring, power line monitoring, UV environmental monitoring, and UV astronomy 4, , 5 6. In the past few years, technological and scientific advances in high Al composition AlGaN and AlN based semiconductor materials have led to a renewed interest in ultraviolet photodetectors, especially solar blind photodetectors (due to the low natural background). Solar-blind detectors allow for a number of unique applications, and the vast majority of the applications listed above take advantage of the solar blind region of the spectrum as illustrated in Figure 2 below.
26
Figure 2. III-Nitride UV photodetectors have a number of civilian and military applications as illustrated above.
Heat sources such as flames, jet engines, or missile plumes emit light throughout the UV portion of the spectrum corresponding to their black-body temperature. These man-made UV sources can easily be detected at wavelengths less than 290 nm due to the non-existence of a terrestrial background signature. The military in particular is interested developing ground an air based solar-blind sensors to detect the UV signature of an active missile plume, and provide early warning and potentially allow for missile tracking and ultimately interception.7,8 A s
olar-blind UV based tracking device would be immune to solar interference and thus capable of tracking a signature across the whole horizon. For non-tracking applications
focused only on threat detection, utilizing a solar-blind detector greatly simplifies the design of the system, eliminating the need to monitor and discriminate against the background.
27
Another important application of UV photodetectors or special interest to the government is covert UV-based non-line-of-sight (NLOS) communications.9 UV-NLOS communications is a secure means to send data using low-power UV sources, and relies on the strong back-scattering of UV and low natural background. The basic operation of such a system is illustrated in Figure 3 below.10 The emitter is pointed towards the sky and the UV light is scattered back towards the earth; due to the short wavelength UV is scattered more strongly than other wavelengths. This makes the system ideal for use where a direct line-of-sight cannot be established, such as in dense terrain or in an Urban-canyon environment when presence of tall cement and iron buildings would make radio communications difficult.
Urban Canyons
Principle of Operation Principle of Operation
Rayleigh Scattering
Source
λ ≤ 280nm Detector
λ ≤ 280nm 10 to 250 m
Squad and Device Communications
Figure 3. Illustrations of the operation of secure UV-based NLOS communication system in a variety of environments.
28
Solar-blind UV has a strong extinction coefficient, due to both high scattering and the high absorption at these wavelengths. This makes eavesdropping on the NLOS
communication’s UV signal very difficult from any significant distance, particularly in the forward direction: this is in marked contrast to conventional RF which can travel thousands of miles or more. The UV signal is almost completely extinguished after a distance of
approximately 250 meters making this a secure covert means of communication. Both the UV detectors and the UV sources necessary for such a system can be realized in the III-Nitride material system making it an ideal choice for the development of a compact secure portable communication system.
Another significant government application of UV detectors (and sources) is biological agent detection. Biological agents could have devastating effects on public health, as the anthrax scare of 2001 made us all too aware.11 There is a significant lag time between a covert attack and the wide-spread appearance of symptoms which makes the general lack of readily available real time detection systems a significant problem. These agents, such as anthrax, smallpox, Marburg virus, Ebola virus, pneumonic plague, and tularemia can, in principle, be simply manufactured and transported in mass quantities, and can cause high rates of mortality if sufficient mitigation procedures are not enacted in a timely manner in place. This makes the development and dissemination of an effective low cost real time detection system a critical weapon in the defense against a bio-terror attack, allowing authorities to time to warn the population, identify the contaminated areas, and enact quarantine procedures before the exposure overwhelms response capabilities.12 The quicker the detection can be made, the less the spread of biological agents will be allowed to spread limiting the quarantine area necessary,
29
and saving more lives, as well as reducing the expenditure of resources necessary to enact a safe and effective cleanup.
Many of the organic and inorganic compounds that make up a bio-agent have absorption lines and/or florescence lines in the UV region of the spectrum. By utilizing the AlGaN material systems to realize detectors and emitters tuned to these strategic wavelengths, it is possible to create a spectral fingerprint which can be used to identify the presence of specific biological-agents in real-time.13,14 Figure 4 shows the absorption and emission spectra of the three most important biological markers, and the role that UV detectors and emitters play in their spectroscopic detection.
340 nm 470 nm NEED:
UV Emitters at λ~280, 340 nm
NEED:
Photodetectors/filters with cutoff λ~300~470 nm
250 300 350 400 450 500 550
Tryptophan
A F
356 nm 278 nm
Normalized Intensity (a.u.)
Wavelength (nm)
300 400 500
A F
303 nm 274 nm
Normalized Intensity (a.u.)
Wavelength (nm)
Tyrosine
Figure 4. UV Absorption and fluorescence of three common biological markers (tyrosine, tryptophan, NADH) illustrating the detector and emitter requirements.
A more civilian application of UV photodetectors is the monitoring of high voltage electrical transmission equipment. Due to exposure to sunlight, airborne pollution, and the
30
weather, the insulators on high voltage transmission equipment degrade with time. In the case of catastrophic failure the high voltage can ionize the air and cause a flash-over where the high voltage arcs across the insulator shorting the system and tripping protection fuses in the
distribution network leading to outages effecting potentially thousands of homes and
businesses.15 However prior to this catastrophic failure occurring, the partial ionization of the air around the insulator leads to the emission of high energy photons, many falling in the solar- blind region of the spectrum. Developing a solar-blind camera would allow for monitoring of this faint UV signature, even during board daylight. In addition, the solar-blind UV portion of the spectrum is ideal for early detection of this coronal discharge and can provide superior performance to thermal imaging.16
Another application of special civilian interest is UV astronomy. Many objects in the sky emit in the visible portion of the spectrum, and many important discoveries have resulted from observation at these wavelengths. However, young stars and stellar remnants (white dwarfs) tend to be much hotter emitting substantial quantity of their radiation in the ultraviolet portion of the spectrum. In addition, many of the important atomic resonance lines are in the UV, or are Doppler shifted into the UV.17 This makes UV astronomy ideal for studying the origins and elemental makeup of the universe. Of particular interest are solar-blind imagers;
while there is no ozone layer in space, objects studied in UV astronomy are often 4 to 8 orders of magnitude brighter in the visible than in the UV, and a high degree of visible-blindness is necessary to discriminate effectively.18
Probably the most promising industrial application for UV photodetectors would be UV flame detection, and combustion monitoring.19 Undetected burner extinction can have deadly consequences if the fuel flow is not stopped immediately and combustible gasses are allowed to
31
build up to unsafe levels. In the late 1950’s primitive visible-blind UV detection tubes were invented20, and industrial furnaces and boilers have been fitted with similar devices ever since.
These tubes rely upon the photoelectric effect to create the electrons necessary to ionize a gas and allow for conduction. They have the advantage of visible-blind operation and faster response than anything that had been invented at that time. However the spectral shape is not ideal for flame detection, and the devices are fragile glass tubes that can burn-out with time.21 In addition to monitoring the presence of known flames, UV flame detectors can be integrated into the fire suppression system as a way of reliably detection fire in environments not suitable to the use of traditional smoke and thermal detectors. UV fire detection is particularly well suited to the detection of alcohol or hydrogen fires where the actual flame can not be easily seen by the human eye.22 The development of a low cost reliable detector tuned to the detection and/or monitor of flames would be an ideal application for III-Nitride detectors.
2.3. UV Detection Technologies
For many decades, the detection of UV light has been accomplished using
photomultiplier tubes (PMTs). These enjoy a high sensitivity to UV photons while being insensitive or “blind” to photons with wavelengths longer than the detector cutoff wavelength.
However, they are fragile vacuum tube devices that require bulky high-voltage power sources to operate. This inherent complexity also makes them relatively expensive. A solid-state alternative to PMTs is silicon-based photodetectors.23 However, Si-based devices are not as robust as AlGaN based photodetectors, and they have considerable sensitivity to photons in the visible and infrared spectral regions, in addition to the ultraviolet portion of the spectrum. The
32
out of band response is commonly addressed by the use of filters, such as a Woods glass optical filter. However, these filters increase the size and weight of the device, and reduce the overall quantum efficiency of the system. It is only in the second half of the 1990’s that wide bandgap III-Nitride semiconductors, AlxGa1-xN in particular, have begun to emerge as the most
promising material systems for such a device, thanks to their exceptional material properties.
III-Nitrides are uniquely suited to the detection of ultraviolet light. The AlGaN material system covers the bulk of the UV portion of the spectrum, allowing tunable cutoffs from 200 nm (6.2 eV) to 365nm (3.4 eV). The addition of indium increases the range even further allowing III-Nitrides to cover most of the UV, including the strategic solar-blind UV, and the entire visible portion of the spectrum, as shown below in Figure 5. No other semiconductor material system can offer such a wide tunable coverage of the ultraviolet and visible portion of the spectrum.
3.0 3.5 4.0 4.5 5.0 5.5 6.0 6.5 7.0 0
1 2 3 4 5 6 7
BN
C
C
MgSe
SiC
ZnO
ZnS
ZnSe CdS
CdSe ZnTe
CdTe GaPAlP
AlSb AlAs
InSb InP
InAs GaSb GaAs
InN GaN
AlN
Bandgap Energy (eV)
Lattice Constant
Solar Blind
(A)
Direct Bandgap IR
IR UV UV
Indirect Bandgap
Figure 5. The quaternary AlInGaN material system is indicated in blue. For comparison other viable semiconductor materials are shown in black.
33
In addition to allowing tunable coverage of the UV and visible portions of the spectrum, III-Nitride based detectors are able to do this with efficiencies that exceed those of other
detector options. The theoretical quantum efficiencies for a number of different photocathode devices, and filtered CCDs are shown below in Figure 6 for comparison to AlGaN detectors. In can clearly be seen that AlGaN offers a significantly higher detective quantum efficiency in the longer wave portion of the spectrum, only being bested in the very short wavelengths
approaching the soft x-rays.
AlGaN AlGaN
Figure 6. Detective Quantum Efficiencies of materials for potential use in UV imaging applications. In the solar blind region of the spectrum AlGaN has a clear advantage over both photocathodes such as Cs2Te , and UV enhanced CCDs
The advantages of a III-Nitride based approach to solar blind UV detection are even more pronounced when one considers imaging applications, as summarized below in Figure 7. The additional size and complexity of a photocathode and micro-channel plate approach to UV
34
imaging (the 2D analog of a PMT) make these devices extremely expensive, bulky, and fragile.
Comparing flittered CCDs to AlGaN based detectors, the efficiency is almost ten times larger, making AlGaN devices much more sensitive.
Photocathode & Microchannel plate
- Based on fragile vacuum tube technology.
- Requires bulky high voltage power supply to opperate.
- Intrinsically less efficient than AlGaN photodetectors at most wavelengths
Filtered Sillicon CCDs
- UV enhanced Silicon CCD - Solar blindness achieved via a
woods filter.
- Suffer from high out-of-band detection,
- Intrinsically less efficient than AlGaN photodetectors
Figure 7. Alternatives to AlGaN for solar blind imaging, and their disadvantages
2.4. Historic Development of III-Nitride Based UV Photodetectors
The development of III-Nitride based ultraviolet photodetectors took off in the early 1990’s. The initial research work focused on GaN-based detectors, and was primarily an offshoot from the technological advances resulting from the drive to develop high quality GaN material for blue light emitting diodes and lasers.24 The few photodetectors that were
demonstrated at the time were simple GaN photoconductors and Schottky photodiodes. During the second half of the 1990’s wide band gap GaN had established itself as a promising material for UV detectors, and soon AlxGa1-xN was being investigated for development of detectors operating over the entire 200~400 nm range. These earliest photodetectors were primarily photoconductors27,28, Schottky metal-semiconductor-metal detectors 35, , , , 37 36 38 39, and Schottky barrier photodiodes37,36. By the late 1990’s research began to focus of the realization of p-i-n
35
photodiodes, , , , , , 53 54 55 56 59 60, because of a number of intrinsic advantages. As an alternative approach to realizing UV p-i-n detectors, the negative electron affinity of GaN was used to make photocathode detectors.47, ,48 49 . (A nice table comparing the performance characteristics of these early detectors can be found the last page of Ref 8).
More recently, research has been geared toward achieving shorter cut-off wavelength p- i-n photodetectors and has especially focused on developing detectors operating in the strategic solar-blind window 57, , , 58 61 62. This push for shorter wavelength has required the use of higher Al-content compounds, and has introduced many technological challenges related to the growth and processing of this wide bandgap material. However, by the early 2000’s, AlxGa1-xN
photodetectors had reached a certain level of maturity, and back-illuminated
photodetectors117,122 were being investigated for potential solar- and visible-blind UV imaging applications142, 143 144 145 146 147 148 149 150, , , , , , , .
The latest stage in III-Nitride based detector development has been the push for the realization of avalanche photodiodes (APDs). A number of groups reported Avalanche multiplication in GaN based devices in the late 90’s, but these early reports were attributed to noisy micro-plasma multiplication.65, 66Latter around 2000, groups discovered that they could suppress the micro-plasmas by adopting smaller device designs.67, , , , , ,68 69 70 71 72 73 The most significant recent APD results have been the discovery of solar-Blind APDs in 2005.74,75 However GaN APD continue to be of interest in parallel with this.25,26
2.4.1. Photoconductors
Photoconductive detectors were the first III-Nitride based UV to be demonstrated, as they are generally the simplest detectors to grow and fabricate. They consist of a simple slab of
36
semiconductor material with two ohmic contacts. After GaN, the first AlxGa1-xN
photoconductive detectors covering the complete range of Al concentrations (0≤x≤1) were rapidly reported.27
By growing single undoped epilayers with thicknesses 0.5~1.5 μm across the entire AlGaN compositional range, UV photoconductors exhibiting sharp cut-off wavelengths from 365 to 200 nm have been realized as shown in Figure 8. These devices were grown by MOCVD on basal plane sapphire substrates; the metal contacts were Ti/Au. The peak
responsivity for x=0.34 (λcutoff ~ 285nm) is about 0.6 A/W. This was the first proof that AlxGa1- xN materials were suitable for solar-blind detector applications.28
200 250 300 350 400
Al
xGa
1-xN
x=0.0 (GaN) x=0.21
x=0.34 x=0.47
x=0.64 x=0.75 x=1.0 (AlN)
Photoresponse (a.u.)
Wavelength (nm)
(00•1) Al2O3
AlxGa1-xN
Figure 8. Photoconductors exhibitinbg sharp cut-off wavelengths covering the entire AlxGa1-xN compositional range: from GaN (365nm) to AlN (200 nm). The inset shows a simplified photoconductor structure.
37
However recently interest in photoconductors has subsided due to the inability to fully resolve the issue of persistent photoconductivity. Persistent photoconductivity in III-Nitrides has been observed in both GaN and AlGaN based photodetectors based on un-doped, and both p-type and n-type photoconductors.29, 30 31 32 33, , , The increased conduction persists long after the photo-excitation is removed, and the effect can easily last for up to several hours. This makes the responsivity of the photoconductor a complex function of the time the sample has been illuminated, or kept in the dark.
Persistent photoconductivity is commonly attributed to defects and dislocations in the material. It is proposed that charges accumulate at the surfaces of the photoconductor and around bulk dislocations. These space charge regions then modulate the effective conduction cross section of the photoconductor. Both the persistent photoconductivity and the gain dependence on optical power have been modeled by considering this light induced band bending due related to defects.34
2.4.2. Schottky Metal-Semiconductor-Metal Detectors
Schottky metal-semiconductor-metal (MSM) photodetectors are also relatively simple photodetectors to realize; III-Nitride based devices were first realized soon after
photoconductors. They generally consist of a single epitaxial layer with two interdigitated Schottky metal contacts deposited on the surface; this creates two back-to-back rectifying junctions. Electron-hole pairs are generated when photons are absorbed near the depletion regions formed at these Schottky junctions, and this lead to a photo-response. The geometry of a typical interdigitated finger device is shown in Figure 9 below.
38
Figure 9. Shows the geometry of a typical interdigitated finger MSM device with a length of 150 μm a finger width of 2 μm and a pitch of 10 μm.
In the case of GaN and AlxGa1-xN materials, these devices are arguably simpler to fabricate because there is no need to highly dope the material and to achieve ohmic contacts.
MSM detectors exhibit all of the desirable attributes of a practical photodetector, such as: high gain, low dark current, high speed, large bandwidth and high sensitivity. However, these devices require an applied bias to operate, and their performance characteristics are dependant on this applied bias, since it changes the volume of the depletion region.
GaN based Schottky MSM devices have been extensively studied.35,36 A typical spectral response from such a detector is shown in Figure 10. A three order of magnitude sharp cut-off at the band-edge of GaN is observed. The high responsivity obtained is indicative of the presence of internal gain in these devices (similar to photoconductors). There have also been reports of detectors with different layer thicknesses and finger geometry which exhibited external quantum efficiencies of up to about 50 % with applied bias in the range of 5 to 20 V
39
and without internal gain.37 These devices also show very low dark currents, as low as 800 fA at a bias of 10 V.
200 300 400 500 600 0.1
1 10 100
Resp on si vi ty ( A /W )
Wavelength (nm)
Figure 10. Shows a typical spectral response of a Schottky based MSM photodetector.
Latter research has focused more on realizing shorter wavelengths, in particular solar- blind, MSM detectors. For example, using an Al0.4Ga0.6N epilayer on sapphire, an external quantum efficiency as high as 49 % (responsivity 107 mA/W) at a 90 V bias has been reported for an MSM detector with peak responsivity wavelength of 272 nm.38
Most recently, back-illuminated solar-blind MSM detectors have been investigated in an effort to move the technology toward focal plane arrays in which the epilayer (front) side of the device needs to be connected to the readout circuitry.39 In the case of these MSM detectors, back-illumination has the advantage of avoiding the blockage of incident photons by the
interdigitated metal contacts, thus enhancing the quantum efficiency. However, because the
40
depletion region (the active region) of the device is located at the epilayer/metal interface (front), the incident photons have to first traverse the substrate and most of the AlxGa1-xN epilayer before reaching the active region. This can be partially circumvented by utilizing a heterostructure in which a larger bandgap epilayer (e.g. AlxGa1-xN) is first grown on the substrate before the active AlyGa1-yN layer with x>y.
Schottky MSM detectors hold some promise for the realization of commercial solar- blind detectors. However these devices require application of a bias, which can be significant when high concentrations of Al are used in solar-blind devices. For this reason and others, fabrication of focal plane arrays seems to have exclusively favored the use of p-i-n based structures over these MSM devices.
2.4.3. Schottky Barrier Photodiodes
Schottky barrier photodiodes have received much less attention than other types of photodetectors40, solar-blind AlxGa1-xN Shottky photodiodes weren’t demonstrated until the year 200041, mainly because of the very rapid interest and progress in Schottky metal- semiconductor-metal detectors. Schottky barrier photodetectors consist of a layer of
semiconductor with two different contacts, one ohmic and one rectifying. Electron-hole pairs are generated when photons are absorbed near the depletion region formed at the Schottky junction. This leads to a photo voltage developing across the two contacts.
The typical spectral response of a Schottky barrier photodetector is shown in Figure 11, with the corresponding device structure shown in the inset. A peak responsivity of 70 mA/W was realized at 272 nm, without applied bias, corresponding to an external quantum efficiency
41
of 32 %. Schottky barrier detectors exhibit good solar-blindness: this example shows a four order of magnitude rejection ratio between the peak response and visible wavelengths.
2 0 0 3 0 0 4 0 0 5 0 0 6 0 0 1 0
-61 0
-51 0
-41 0
-31 0
-21 0
-1Responsivity (A/W)
W a v e le n g th (n m )
S p e c tr a l R e sp o n s iv ity
3 0 0 0 Å A lG a N 1 μ m A lG a N: S i
S a p p h ire
T i/A u N i/A u
5 0 0 0 Å G a N :S i 2 μ m
D e v ic e S tr u c tu r e
G a N :S i
Figure 11. Shows the typical spectral response of a Schottky barrier photodetector. The corresponding device structure is shown in the inset.
2.4.4. Photocathodes
A less traditional approach to realizing UV detectors that has been studied in parallel with the other approaches it the fabrication of photocathodes. Photocathodes rely on the negative electron affinity of GaN42,43 to convert photons into photo-electrons that can then be directly read out, or amplified with an electron multiplication cascade. For imaging applications this would consist of the photocathode and an electron bombardment CCD for direct readout or the insertion of a micro-channel plate for electron signal amplification before readout.
