`
`Herbert Paul Maruska
`
`November, 1973
`
`TCL 1028, Page 1
`
`
`
`GALL I UM NITRIDE LIGHT-EMITTI NG DIODES
`
`A DISSERTATION
`
`SUEt-fiTTED TO THE DEPARTMENT OF MATERIALS SCIENCE AND ENGINEERING
`
`AND THE COMMITTEE ON GRADUATE STUDIES
`
`OF STANFORD UNIVERSITY
`
`IN PARTIAL FULFILLMENT OF THE REQUIREMENTS
`
`FOR THE DEGREE OF
`
`DOCTOR OF PHILOSOPHY
`
`By
`
`Herbert Paul Maruska
`
`November 1973
`
`TCL 1028, Page 2
`
`I
`
`~. ,,
`
`r L
`
`
`
`r
`
`I certify that I have read this thesis and that in my
`opinion it is fully adequate, in scope and quality, as
`a dissertation for the degree of Doctor of Philosophy.
`
`b~k~~a~ (Principal Adviser)
`
`I certify that I have read this thesis and that in my
`opinion it is fully adequate, in scope and quality , as
`a dissertation for the degree of Doctor of Philosophy.
`
`I certify that I have read this thesis and that in my
`opinion it is fully adequate, in scope and quality, as
`a dissertation for the degree of Doctor of Philosophy .
`
`/2~ J-6
`
`/_ &~~
`
`· (Electrical Engineering)
`
`I certify that I have read this thesis and that in my
`opinion it is fully adequate, in scope and quality , as
`a dissertation for the degree of Doctor of Philosophy.
`
`Approved for the University Committee
`on Graduate Studies:
`
`~~~~~ f U-u:w
`
`Dean of Graduate Studies
`
`ii
`
`TCL 1028, Page 3
`
`
`
`r
`
`l
`
`To Jacques I. Pankove , whose dedicated
`
`interest in science has been a source of
`
`inspiration to me.
`
`i i i
`
`TCL 1028, Page 4
`
`
`
`r
`
`ACKNOWLEDGMENTS
`
`I would like to thank the following individuals for their contri(cid:173)
`
`butions to this dissertation:
`
`Professor David A. Stevenson, principal advisor, for guidance,
`
`helpful discussions, and continuing interest.
`
`Professor Richard H. Bube, Professor James F. Gibbons , and Professor
`
`Gerald L. Pearson, members of the thesis reading commit tee, for helpful
`
`suggestions and discussions.
`
`Dr. Jacques I. Pankove of RCA Laboratories for starting me on a
`
`career in scientific research and for his lasting friendship.
`
`Dr. Robert A. Burmeister, Jr. and Dr. Egan E. Loebner of Hewlett
`
`Packard for a critical review of the manuscript.
`
`Dr. Wally Rhines of Texas Instruments for experimental assistance
`
`and many creative and stimulating discussions while he was at Stanford,
`
`and continuing friendship .
`
`Dr. Troy W, Rarbee and Professor Robert H. Huggins for valuable
`
`suggestions and discussions.
`
`Mr. William L. Larson for much appreciated experimental assistance
`
`and good friendship.
`
`Mr. Steve Chiang, Dr. Alan L. Fahrenbruch , Mr . Bruce E. Liebert,
`
`Mr. John McKenzie, and Mr . Yves Verhelle for helpful discussions and
`
`friendship .
`
`Mr . Perfecto Mary for preparing the drawings and illustrations.
`
`Mrs. Valerie Faulkenburg for typing the original and final
`
`manuscripts.
`
`TCL 1028, Page 5
`
`
`
`1
`
`The assistance of Mr. Loren Anderson, Miss Rosemary Koch, Mrs.
`
`Marta de Rojas and the entire technical and office staffs of the Center
`
`for Materials Research is gratefully acknowledged.
`
`Two years of fellowship support from RCA Laboratories is also
`
`gratefully acknowledged.
`
`And I would like to thank my wife, Claire, for all the good times
`
`we had in California.
`
`This research was supported by the Advanced Research Projects
`
`Agency under Grant No. DAHC15 71-G-6.
`
`v
`
`TCL 1028, Page 6
`
`
`
`TABLE OF CONTENTS
`
`Chapter
`
`1.
`
`LIGHT-EMITTING DIODES •
`
`1.1
`
`Introduction .
`
`1.2 Classification and Basis of Operation
`of LED's .
`,
`,
`,
`
`1.3 LED Efficiencies
`
`1.4 Examples of LED Materials
`
`1.5 Purpose of This Work • . •
`
`2.
`
`SYNTHESIS TECHNIQUES AND PROPERTIES OF
`GALLIUM NITRIDE
`.
`.
`.
`.
`. •
`. • •
`. •
`
`.
`
`2.2 Doping
`
`2. 1 Synthesis and Growth Techniques
`. .
`.
`2.3 Optical Properties
`
`2.4 Electrical Properties
`2.5 Electroluminescence . . . .
`EXPERIMENTAL TECHNIQUES •
`
`3.
`
`3.1 Open- Flow Vapor Growth System
`
`3. 2 Doping Procedure .
`
`.
`
`3.3 Point-Contact LED ' s
`
`3. 4
`
`Improved Large-Area He tal Contacts
`
`3.5 Heasurement of Wavelength and Intensit y
`of Luminescence
`.
`.
`.
`.
`.
`
`3. 6 Elec tr ical Chara cteristi cs
`
`3.7 Scanning Electron Microscope
`
`3. 8 Trans missio n Electron Microscope
`
`3.9 Proton Bombardment and Proton- Assisted Diffusion
`
`1
`
`1
`
`2
`
`7
`
`10
`
`23
`
`25
`
`25
`
`31
`
`32
`
`36
`
`39
`
`41
`
`41
`
`4 7
`
`53
`
`54
`
`57
`
`63
`
`67
`
`70
`
`71
`
`vi
`
`TCL 1028, Page 7
`
`
`
`TABLE OF CONTENTS (Contd)
`
`Chapter
`
`4.
`
`EXPERU1ENTAL RESULTS
`
`4 .1 Growth and Fabrication of GaN Light-
`Emitting Diodes
`
`4.2 Emission Spectra of GaN:Mg Light(cid:173)
`Emitting Diodes
`.
`.
`. •
`.
`
`4.3 Electrical Charac teristics •
`
`4.4 Temperatu re Dependence of Luminescence
`and Electrical Characteristics •
`
`4.5 Growth and Surface Morphology of GaN
`
`4 . 6 Pattern of Light Emission
`
`4 . 7 Characteristics of the Insulating
`(Mg- doped) Region
`. •
`. • •
`.
`
`4 . 8 Electrical Po t ential Distribution
`
`4.9 Diode Capacitance
`
`73
`
`74
`
`75
`
`84
`
`89
`
`96
`
`103
`
`117
`
`130
`
`134
`
`4.10 Proton Bombardment and Proton-Assisted
`Dif f usion
`
`141 -.·
`
`5.
`
`DISCUSSION
`
`5 .1 Basic Mechanisms fo r Diode Operation .
`
`5.2 Nechanism for GaN m-i- n Diode Electrical
`Characteristics
`
`5.3 Nechanisms fo r Electroluminescence
`
`5.4
`
`Impact Ionization in GaN m- i-n Diodes
`
`5.5 Nature of the Potential Barrier
`
`5 .6 Model for GaN Light-Emitting Diodes
`
`6 .
`
`CONCLUSIONS . . . . .• .
`
`. . • •
`
`143
`
`143
`
`154
`
`160
`
`171
`
`180
`
`185
`
`193
`
`vii
`
`TCL 1028, Page 8
`
`
`
`LI ST OF FIGURES
`
`Number
`
`Figure 1.1
`
`Spectral sensi tivity of the human eye .
`
`Figure 1.2
`
`Best reported values of luminous power con(cid:173)
`version for light- emitting diodes.
`
`Figure 1.3
`
`Chromatic it y diagram.
`
`Figure 1. 4
`
`The color matching functions V(x) , V(y),
`and V ( z) , (red, green, and blue) •
`
`Figure 2.1
`
`Equilibrium vapor pressure of N2 over
`GaN(s) -Ga( £).
`
`Figure 3.1
`
`Open-Flow Vapor Growth System for pre(cid:173)
`paring gallium nitride.
`
`Figure 3.2
`
`Quartz sample holder.
`
`Figure 3.3
`
`Sys tem for positioning dopant crucible
`in sidearm of vapo r growth apparat us.
`
`Figure 3 . 4
`
`Graphite dopant crucible.
`
`Figure 3.5
`
`Gal lium nitride light-emitting diode
`attached to a glass slide.
`
`Figure 3.6
`
`Schematic diagram of a gallium nitride light(cid:173)
`emitting diode on a T0-5 trans istor header .
`
`Figure 3.7a Experimental arrangement for measuring t he
`wavelength and intensity of elec trolumi(cid:173)
`nescence (schematic).
`
`Figure 3.7b Experimental arrangement for measuring t he
`wavelength and intensity of photol umines(cid:173)
`cence (schematic).
`
`Figure 3.8
`
`Spectral response of the S-20 photocathode.
`
`Figure 3.9
`
`System for measuring 1-V characteristics
`(schematic) .
`
`Figure 3.10 System for simultaneous measurement of power
`efficiency and I - V characteristics of a light(cid:173)
`emitting diode (schematic) .
`
`11
`
`15
`
`1 7
`
`18
`
`30
`
`43
`
`49
`
`52
`
`56
`
`58
`
`59
`
`60
`
`62
`
`64
`
`66
`
`viii
`
`TCL 1028, Page 9
`
`
`
`LIST OF FIGURES (Contd)
`
`Number
`
`Page
`
`Figu r e 3.11 Schematic diagram of the scanning
`electron microscope .
`
`Figure 4. 1
`
`Elec t ro1uminescence spectrum of point-cont ac t
`GaN :Mg light- emitting diode (sample #6 · 8 · 72).
`
`Figure 4.2 Electroluminescence spectrum of GaN:Mg m-i-n
`diode, sample #6-28·72 (forward bias).
`
`Fi gure 4. 3 Elect roluminescence spectrum of typical
`GaN:Mg m- i - n diode (forward bias).
`
`Figure 4 . 4 Dependence of t he peak wavelength of the emission
`on app l ied voltage (forward bias).
`
`Figur e 4.5 Typical reverse-bias electroluminescence
`spectrum for GaN:Mg m-i-n diodes .
`
`Figure 4 . 6 Dependence of the peak wavelength of the emi ssion
`on applied voltage (reverse bias) .
`
`Figure 4 . 7
`
`Photoluminescence spectrum of Mg-doped GaN.
`
`Figure 4.8 Dependence of light power output on electrical
`power input for GaN:Mg ligh t -emitti ng diodes .
`
`Figure 4. 9 External quantum efficiency of GaN:Mg light(cid:173)
`emitting diodes.
`
`Figure 4.10 Typ ical diode I-V characteristic .
`
`Figure 4.11 Log- log plot of diode I - V characteristic
`(sample #7 ·7 · 72) .
`
`Figure 4.12 Luminous intensity as a function of
`applied voltage .
`
`Figure 4 . 13 Temperature dependencies of diode current and
`luminous intensity at constant vo ltage .
`
`Figure 4.14 Temperature dependencies of diode current and
`luminous intensity at various constant voltages .
`
`Figure 4 . 15 Temperature dependence of photoluminescence
`i ntensity.
`
`ix
`
`68
`
`76
`
`78
`
`79
`
`81
`
`82
`
`83
`
`85
`
`86
`
`87
`
`88
`
`90
`
`91
`
`93
`
`94
`
`95
`
`!·
`
`.
`l ,.
`TCL 1028, Page 10
`I I!
`
`
`
`LIST OF FIGURES (Contd)
`
`Number
`
`Figure 4.16 Hexagonal islands of GaN nucleated on (0001)-
`sapphire .
`
`Figure 4.17 Hexagonal islands of GaN, which have coalesced
`to form a continuous film on the left side of
`the picture. (0001)-orientation.
`
`Figure 4.18 Surface of gallium nitride sample #5 ·19·72,
`grown on (ll02)-sapphire.
`
`Figure 4 . 19 Typical facetted surf ace developed in GaN
`grown on (ll02)-sapryhire .
`
`Figure 4.19A Gallium nitride films on quartz subst rates ,
`showing crystallite s ize at various distances
`downstream from the ammonia inlet:
`(a) 2";
`(b) 3"; (c) 4". Magnifi cation 200X.
`
`Figure 4.20 Laue pattern of (1013) - GaN (grown in (ll0 2)-
`sapphire substrate) .
`
`Figure 4 . 21 Gallium nitride light-emitting diode with
`external illumination. The (ll02) sapphire
`substrate is remov ed, exposing the n- GaN
`layer . Magnification 28X.
`
`Fi gure 4.22 Light spot pattern observed during f orward
`biasing of the diode shown in Figure 4 . 21 .
`Magnification 28X .
`
`Figure 4 . 23 Light spot pattern of forward- biased diode
`at 200X magnification .
`
`Figure 4.24 Same diodes as in Figure 4.23 with external
`illumination and l ight spot pattern super-
`imposed (200X).
`(10l3)-orientation .
`
`Figure 4.25 Gal l ium nitride light- emitting diode
`#12 · 7· 72 with external illumination .
`(10l3) - orientation .
`
`Figure 4. 26 Gallium nitr ide l ight-emitting diode
`#12· 7·72 (cf. Figure 4 . 25) with forward
`bias , illustrating l ight spot pattern.
`
`97
`
`99
`
`100
`
`101
`
`102
`
`104
`
`105
`
`107
`
`108
`
`109
`
`111
`
`112
`
`X
`
`TCL 1028, Page 11
`
`
`
`LIST OF FIGURES (Contd)
`
`Number
`
`Figure 4.27 Diode #12 •7·72 (cf. Figure 4.25) with
`reverse bias, illustrating different set of
`l igh t spots appearing with reverse bias than
`forward bias.
`
`Figure 4.28 Diode mounted perpendicular to a glass slide
`showing sapphire substrate, CaN, and metal
`contact.
`
`Figure 4.29 Lightspotsobserved when diode shown in
`Figure 4.28 was forward biased. The light
`spots occur within the CaN region.
`Magnification SOOX.
`
`Figure 4.30 Perpendicular view of diode #12 •13·72, as
`seen in the scanning electron microscope,
`The entire device (including the (1102)(cid:173)
`sapphire substrate) and the two metal
`contacts (to the n- and i-layers) are
`visible . Magnification 25X.
`
`Figure 4.31 Scanning electron micrograph of diode
`#12 •13• 72 at lOOX magnification--no bias.
`(ll02)-oriented sapphire substrate.
`
`Figure 4.32 Scanning electron micrograph of diode
`#12 ·13·72 at lOOX magnification with
`forward bi as (+22 volts).
`
`Figure 4.33 Scanning electron micrograph of diode
`#12 · 13 · 72 at lOOX magnification with
`reverse bias (-22 volts) .
`
`Figure 4.34 Scanning electron mi crograph of diode
`#12·13·72 at SOOX magnification with
`reverse bias. The ins ula ting (Mg-doped)
`Ga~ layer can be clearly distinguished.
`(1102)-oriented sapphire substrate.
`
`Figure 4.35 Scanning electron micro graph of diode
`#12·13•72 at lOOOX magnification with
`no bias.
`
`Page
`
`113
`
`115
`
`116
`
`120
`
`121
`
`122
`
`123
`
`124
`
`126
`
`xi
`
`TCL 1028, Page 12
`
`
`
`LIST OF FIGURES (Contd)
`
`Number
`
`Figure 4.36 Scanning electron micrograph of diode
`#12 · 13·72 at lOOOX magnification with
`forward bias. The bias was removed
`during part of the scan, as indicated.
`
`Fi~ure 4.37 Scanning electron micrograph of diode
`#12·13 · 72 at lOOOX magnification with
`reverse bias . The bias was removed
`during part of the scan, as indicated.
`
`Figure 4.38 Transmission electr.on micrograph of
`insulating (Mg-doped) region of a CaN
`LED. Notice no evidence of precipitates
`or dislocations. Magnification 240 , 000X.
`
`Figure 4.39 Superimposed zero bias and forward bias
`line scan traces and SEM photograph of a
`gallium nitride light-emitting diode.
`Magnification lOOOX.
`
`Figure 4.40 Superimposed zero bias and forward bias
`line scan traces and SEM photograph of a
`gallium nitride light-emitting diode.
`Magnification 2500X .
`
`Figure 4.41 Superimposed zero bias and reverse bias
`(-6 , -12, and -18 volts) line scan traces
`and SEM photograph of a gallium nitride
`light-emitting diode. Magnification 2500X.
`
`Fig~re 4.42 Electric potential distribution across a
`gallium nitride light- emitting diode when
`forward bias is applied.
`
`Figure 4.43 Electric potential distribution across a
`gallium nitride light-emitting diode when
`reverse bias is applied.
`
`Figure 4.44 The capacitance and conductance as functions
`of applied voltage for a typical CaN diode .
`Also included is the quantity, l /C2 , as a
`function of voltage.
`
`Figure 4.45 Current-voltage characteristic of a GaN
`m- i - n diode prepared by proton bombardment.
`
`127
`
`128
`
`129
`
`131
`
`132
`
`133
`
`135
`
`136
`
`138
`
`140
`
`xii
`
`TCL 1028, Page 13
`
`
`
`LIST OF FIGURES (Contd)
`
`Number
`
`Figure 5.1
`
`Figure 5.2
`
`Fowler-Nordheim plot of GaN:Mg light(cid:173)
`emitting diode current-voltage character(cid:173)
`istic. Cf. Figure 4.11,
`
`Fowler-Nordheim plot of GaN:Mg light(cid:173)
`emitting diode current-voltage character(cid:173)
`istics at various temperatures.
`Cf. Figure 4.14.
`
`Figure 5.3
`
`Plot of log I versus log V for GaN:Zn
`light-emittinB diode (after PANKOVE,
`1972, Figure 8).
`
`Figure 5.4
`
`The I-V characteristic of the GaN:Zn
`light-emitting diode shown in Figure 5.3,
`replotted to the Fowler-Nordheim Equation.
`
`Figure 5.5
`
`Log of electroluminescence intensity versus
`(voltage)-l/ 2. Cf. Figure 4.12.
`
`Figure 5.6
`
`Log of forward and reverse bias electro(cid:173)
`-1/2
`luminescence intensity versus (voltage)
`•
`
`Figure 5. 7
`
`Log of electroluminescence intensity versus
`-1/2
`(voltage)
`at various temperatures.
`Cf. Figure 4.14.
`
`Figure 5.8
`
`Log of luminous intensity versus log of
`diode current at various temperatures.
`Cf. Figure 4.14.
`
`Figure 5.9 Carrier multiplication factor M as a
`function of applied voltage for a GaN
`light-emitting diode.
`
`Figure 5.10 Plot of (external quantum efficiency) x
`(voltage)-l/ 2 versus (voltage)-l/ 2 •
`
`Figure 5.11 Charge concentrations, electric fields, and
`electric potentials at a p-n junction. NA,
`Nn are respectively the concentrations of
`acceptors and donors; FA, FD, and F are
`respectively the electric field due to the
`
`xiii
`
`155
`
`156
`
`157
`
`158
`
`163
`
`164
`
`165
`
`166
`
`176
`
`179
`
`TCL 1028, Page 14
`
`
`
`LIST OF FIGURES (Contd)
`
`Number
`
`i onized acceptors, elec tric field due to the
`ionized donors, and total electric field;
`~A' ¢D, and ¢ are respectively the potential
`due to the ionized a cceptors, potential due
`to the ionized donors, and total potential;
`and ¢B is the barrier potential .
`
`Figure 5.12 Schematic of processes responsible for
`electroluminescence in GaN m-i-n diodes.
`The situation for forward bias is
`illustrated.
`
`Figure 5 . 13 Growth and device operation of a gallium
`nitride light- emitting diode:
`(a)
`Nucleation of GaN islands on a (ll02)
`sapphire substrate; (b) Growth of GaN
`islands, until they meet; (c) Growth of
`continuous GaN film after the islands
`have met, indicating the position of the
`sub- grain boundaries; (d) Complete GaN
`m-i-n diode , showing the structural origin
`of the e lectroluminescence with forward
`bias.
`
`182
`
`187
`
`189
`
`xiv
`
`TCL 1028, Page 15
`
`
`
`LIST OF SYMBOLS
`
`constant; lattice parameter (A); activity
`area (cm2 ); constant
`
`0
`
`constant
`
`barrier height (eV)
`
`constant; lattice parameter (A)
`
`0
`
`capacitance (pF)
`
`thickness of insulator (em)
`
`energy (eV)
`
`activation energy (eV)
`
`band gap energy (eV)
`
`minimum energy for impact ionization (eV)
`
`electric field (volts/em)
`
`luminous power (lumens)
`
`Gibbs free energy (Kcal)
`
`standard Gibbs free energy (Kcal)
`
`Planck's constant(4.14 x 10-lS eV-sec)
`
`insulating
`
`current (amps)
`
`2
`current density (amp/em )
`Boltzmann constant (0.864 x 10-4 eV/°K)
`
`equilibrium constant
`
`a
`
`A
`
`b
`
`B
`
`c
`
`C
`
`d
`
`E
`
`E
`a
`
`E
`g
`
`E
`m
`
`F
`
`F1
`G
`
`Go
`
`h
`
`i
`
`I
`
`J
`
`k
`
`K
`
`length (em); free path (em); dimensionless parameter
`characterizing trap distribution
`
`mean free path (em)
`
`m
`
`metal
`
`XV
`
`TCL 1028, Page 16
`
`
`
`,
`
`LIST OF SYMBOLS (Contd)
`
`mass of electron (9 . 1 x l0-31 Kg)
`
`effective mass of electron (Kg)
`electron carrier density (cm- 3); numerical constant
`effective charge dens ity in depletion region (cm-3)
`-3
`acceptor density (em
`)
`donor density (cm- 3)
`
`-3
`luminescent center density (em
`)
`hole carrier density (cm- 3)
`
`light power output (wa tts); ionization probability ; pressure
`
`-19
`electronic charge (1 . 60 x 10
`
`coul)
`
`resistance (0); gas constant (1 . 987 cal/°K-mole)
`
`effective electron temperature (°K)
`
`bias voltage (volts)
`
`eye sensitivity function
`
`concentration of positively charged vacancies on the
`nitrogen lattice
`
`width of depletion region (em)
`
`distance (em)
`
`-1
`absorption coefficient (em
`)
`
`permittivity (farads/m)
`
`-12
`permittivit y of free space (8 . 85 x 10
`
`farads/m)
`
`efficiency
`
`m
`
`m*
`
`n
`
`N
`
`p
`
`p
`
`q
`
`R
`
`T
`
`T e
`v
`
`V(.\)
`
`[V. ]
`N
`
`w
`
`X
`
`a
`
`E
`0
`
`n
`
`llEQ
`
`external quantum efficiency
`
`radiative efficiency of recombination
`
`xvi
`
`TCL 1028, Page 17
`
`:l
`
`i I
`
`
`
`J.l
`
`\1
`
`a
`
`1'
`
`LIST OF SYMBOLS (Contd)
`
`0
`
`wavelength (A)
`mobility (cm2/V-sec)
`frequency (sec-1 )
`
`2
`-1
`conductivity en-em
`); cross-section (em)
`
`lifetime
`
`electric potential (volts)
`
`barrier potential (volts)
`
`xvii
`
`TCL 1028, Page 18
`
`
`
`ABSTRACT
`
`The synthesis and characterization of hetero-epitaxial gallium
`
`nitride (GaN) films were undertaken with particular reference to the
`
`phenomenon of light emission. Gallium nitride was grown by the chemical
`
`vapor deposition technique using sapphire substrates. Techniques for
`
`characterization included optical, scanning electron, and transmission
`
`electron microscopy and measurements of various electrical and optical
`
`properties of the films. The thin films of GaN were doped during
`
`growth with zinc and magnesium to form n-i junctions. Such material
`
`provided the basis for the fabrication of m-i-n light-emitting diodes,
`
`which emitted light in the high-energy violet region of the visible
`
`spectrum with Mg doping and green light with Zn doping.
`
`Undoped GaN laye rs were ordinarily n-type with carrier concentra-
`
`18
`tions in excess of 10
`
`-3
`em
`
`Insulating GaN was produced through
`
`compensation by doping with the acceptor impurities Zn and Mg. Light-
`
`emitting diodes were prepared by growing onto a sapphire s ubstrate a
`
`film of highly conduc ting undoped GaN (n-layer), followed by a film of
`
`ins ulating GaN (i- layer) , and then placing a metal contact over the
`
`surface (m-layer). Such a struc ture f orms an m-i-n diode, which con-
`
`tains two electrical junctions, m-i and i-n. Most of the diodes studied
`
`had a Mg-doped i-layer, and such diodes exhibited violet electro-
`
`luminescence which peaked at 2.9 eV with a 400 meV halfwidth. External
`
`quan t um efficiencies of about 0.005% were obtained in these diodes
`
`with a bias of 20 volts.
`
`The l ight is emitted from the diodes in the form of small (<10 ~
`
`xviii
`
`TCL 1028, Page 19
`
`
`
`1 i l
`
`diameter), discrete spots. These light spots are generated at the
`
`junction in the diode which forms the cathode. Although Laue patterns
`
`indicate that the GaN films are single-crystalline, the surfaces of the
`
`layers show a strongly facetted structure, and the positions of the
`
`light spots correspond to these facets. Gallium nitride is nucleated
`
`in the form of a large number of individual islands which subsequently
`
`coalesce to form a continuous film. The coalesced islands are slightly
`
`misaligned, resulting in the formation of subgrain boundaries. No
`
`evidence for precipitates or dislocations within the individual cells
`
`was found.
`
`The electric potential distribution across the m-i-n diodes was
`
`obtained through the voltage-contrast capability of the scanning
`
`electron microscope. Potential gradients and electric fields of about
`105 volts/em are developed at the m-i and i-n junctions, but not
`
`throughout the bulk of the i-layer. These observations, coupled with
`
`a study of the diode current-voltage characteristics, including their
`
`temperature dependence, lead to a model of current control in GaN
`
`diodes by tunnelling through a potential barrier. The I-V character-
`
`istics follow the Fowler-Nordheim field-emission equation.
`
`It is proposed that impact ionization, occurring at the subgrain
`
`boundaries in the GaN films, is respons ible for the electroluminescence.
`
`High-energy electrons emp t y the luminescent centers by collisions, and
`
`the subsequent recombination of electrons in thermal equilibrium in the
`
`conduction band with these centers results in the generation of light.
`
`This model is consistent with photomultiplication measurements and with
`
`the voltage dependence of the external quantum efficiency. The models
`TCL 1028, Page 20
`
`xix
`
`
`
`developed for the diode electrical properties and for the electro-
`
`luminescence are useful i n understanding the characteristics of m-i-n
`
`junctions.
`
`I
`
`1
`
`f.
`
`i
`~
`t
`
`TCL 1028, Page 21
`
`
`
`~r
`
`r
`!
`
`I '
`
`\
`
`'Let there be light,'
`11God said ,
`and there was light. God saw t hat t he
`light was good , and God divided light
`f rom darkness . 11
`
`Gn 1: 3, 4
`
`xxi
`
`TCL 1028, Page 22
`
`
`
`PLATE I. Forward biased GaN:Mg light-emitting diode.
`
`xxii
`
`TCL 1028, Page 23
`
`
`
`PLATE II. Reverse biased GaN:Hg light-emitting diode .
`
`xxiii
`
`TCL 1028, Page 24
`
`
`
`.. ,. l
`
`Chapter 1
`
`LIGHT-EMITTING DIODES
`
`1.1 Introduction
`
`Light-emitting diodes (LED's) are devices that emit electromagnetic
`
`radiation in response to an applied electric field . Light emission from
`
`a material due to an electric field is termed electroluminescence. The
`
`light involved with the process of electroluminescence is not determined
`
`simply by the temperature of the material, as is the case with incan-
`
`descence (black body radiation). Applications for LED's are found in
`
`the areas of indicator lights, alpha-numeric displays, picture displays
`
`(s olid-state color television) and IR sources. Commercially available
`
`red LED's are now being used as numerals in calculators and wrist-
`
`watches, and in place of panel meters in various types of electronic
`
`equipment. For most application, then, it is necessary that the emitted
`
`light fall in the visible portion of the electromagnetic spectrum
`
`0
`
`0
`
`(4000 A to 7000 A), s ince the purpose of the devices is to transmit
`
`information from instruments to people.
`
`Solid-state semiconductor materials are used for the manufacture of
`
`various light-emitting diodes. They feature many attractive qualities
`
`in contrast to incandescent light s ources:
`
`reliability, long life, low
`
`cost, small size, and low power requirements and operating voltages.
`
`The last three items are particularly important for compact battery
`
`operated equipment. The operating characteristics of semiconductor
`
`LED' s are compatible with silicon integrated circuits, which makes them
`
`espec ially suitable for visual dis plays in computer systems .
`
`- 1-
`
`TCL 1028, Page 25
`
`
`
`1.2 Classification and Basis of Operation of LED's
`
`It is possible to distinguish five classes of semiconductor light-
`
`emitting diodes. These are,
`
`(1) p-n homojunction diodes
`
`(2) p-n heterojunction diodes
`
`(3) Schottky barrier and m-i-s diodes
`
`(4) electroluminescent cells (powders dispersed in organic
`
`binders)
`
`(5)
`
`infrared emitting diodes coated with phosphors that emit
`
`visible ligh t ("up-converters").
`
`The first class of diodes involves a single material, part of which is
`
`doped p-type and the other part of which is doped n-type to form a p-n
`
`junction. The n-type side of the diode is prepared so as to contain an
`
`excess of electrons, while the p-type side has a deficit of electrons
`
`(that is, an excess of holes). Excess electrons are introduced into a
`
`semiconductor by substituting in place of a normal lattice constituent
`
`an element (dopant) which contains one more than the proper number of
`
`bonding electrons. Holes occur when the substituted dopant lacks one
`
`from the proper number of bonding electrons. When a potential differ(cid:173)
`
`ence is applied to such a diode, electrons are injected from the n-type
`
`region across the junction i nto the p-type region, where they can
`
`recombine with holes. Conversely, holes can be injected into the n-side
`
`with subsequent recombination with electrons; usually one process is
`
`dominant in a given semiconductor. The carriers which have been
`
`inj ected into a region in which they were originally scarce are called
`
`minority carriers. The carriers present in excess are the majority
`
`-2-
`
`TCL 1028, Page 26
`
`
`
`carriers.
`
`The excess electrons are situated in the semiconductor's conduction
`
`band in the n-type region, while the holes reside in the valence band in
`
`the p-type region. The valence and conduction bands are separated by a
`
`forbidden energy region, called the "energy gap" or "band gap" of the
`
`material. This is just the energy required to move an electron from
`
`the valence band (where a hole is left behind) to the conduction band.
`
`Thus, for example , if electrons are injected from the conduction band
`
`on the n-side to the conduction band on the p-side, under the influence
`
`of an applied potential difference, and they recombine with holes in
`
`the valence band, the total energy difference involved is just the
`
`" band gap" energy . This energy of recombination can be released from
`
`the diode in the form of photons . All of the photons emitted by a
`
`light-emitting diode wi ll not possess exactly the same energy because
`
`the electrons and holes which recombine are distributed in energy in
`
`their respective bands. The energy distributions of the electrons and
`
`holes are governed by Fermi-Dirac statistics, but if the semiconductor
`
`is not degenerately doped, the distributions in the valence and con(cid:173)
`
`duction bands may be approximated by Maxwell-Boltzmann statistics .
`
`Since the carriers are distributed in energy in this way, a plot of
`
`number of photons emitted vs . energy has the form of a bell-shaped
`
`curve .
`
`In many cases the electrons and holes recombine at a localized
`
`energy state within the forbidden energy gap . Such recombination sites
`
`are due to imperfections in the crystal, including the purp osely intro(cid:173)
`
`duced dopant atoms.
`
`In such a case the photon energy will be the
`
`-3-
`
`TCL 1028, Page 27
`
`
`
`energy separation between this level and one of the bands. Notice that
`
`two photons of different energies adding up to the band gap energy could
`
`be emitted, but in many situations of interest the localized state is
`
`quite close (a few tenths of an electron-volt) from either the valence
`
`or the conduction band, so that the energy released is approximately
`
`the band gap energy of the semiconductor.
`
`In any case the band gap
`
`determines the highest frequency range of light which may be emitted.
`
`In order for the light to be visible to the human eye, a minimum band
`
`gap of 1.8 eV is required. The largest band gap that will produce
`
`visible light is about 3.1 eV, although materials with still larger
`
`band gaps may contain appropriate recombination centers to provide
`
`visible light.
`
`In this first class of light-emitting diodes, then, the crystal is
`
`homogeneous throughout except for the change in dopant. The problems
`
`associated with joining dissimilar materials, such as interface states
`
`acting as electronic traps, dislocations, and lattice strains, are
`
`avoided. Various disruptions of the crystalline perfection usually
`
`decrease the number of photons emitted rather than just providing
`
`recombination centers, and the recombination energy is dissipated non(cid:173)
`
`radiatively at the imperfections.
`
`However, it is not possible to form p-n junctions in all materials.
`
`There are substances which exhibit excellent luminescence when stimu(cid:173)
`
`lated with external light sources (photoluminescence) or when bombarded
`
`with electron beams (cathodoluminescence), and therefore appear to be
`
`attractive candidates for electroluminescence. This is especially true
`
`if their band gaps fall in the visible region of the spectrum, However,
`
`-4-
`
`TCL 1028, Page 28
`
`
`
`r
`
`I
`
`only one conductivity type, which, depending on the compound, is either
`
`nor p , _but not both , can be achieved by doping .
`
`In the absence of a
`
`p-n homojunction, other schemes must be devised for creating the
`
`minority carriers necessary for recombination. Such approaches are
`
`involved in diodes of classes 2, 3, and 4.
`
`In class 2 diodes, different
`
`materials are used on the two sides of a p-n junction, forming a hetero-
`
`junction. The semiconductors involved in such an arrangement are one
`
`which can only be doped p-type joined to another which can only be
`
`doped n-type.
`
`In most cases, the energy gaps are not the same in the
`
`two substances, leading to a further complication beyond the lattice
`
`strains involved:
`
`the potential barriers that must be surmounted at
`
`the junction will be different for electrons and holes. Few sets of
`
`materials fit into this heterojunction scheme, because the two compounds
`
`selected should be comple t ely mutually soluble.
`
`If they are not, a two
`
`phase region with poor electrical properties can result in the junction
`
`area.
`
`It is also sometimes possible to obtain electroluminescence from
`
`materials in which one cannot form p-n junctions by the use of Schottky
`
`barriers or m-i-s structures (type 3 didoes) . A Schott ky barrier is
`
`formed at a non- ohmic (i . e., rectifying) metal contact to a semi-
`
`conductor , and light can be generated in the high- field region occurring
`
`at this contact.
`
`In them-i-s (metal-insulator-semiconductor) structure,
`
`an insulating layer, either of the semiconductor itself or some other
`
`substance , is interposed between the metal contact and the semiconductor .
`
`Often such an insulating layer is inadvertantly f ormed when fabricating
`
`a Schottky barrier if the metal used for the contact diffuses into the
`
`-5-
`
`TCL 1028, Page 29
`
`
`
`semiconductor, rendering it insulating (by compensating the dopant
`
`already in it).
`
`In some cases it is possible to inject minority carriers from the
`
`metal (~vhich contains large quantities of both electrons and holes near
`
`its Fermi level) into the high-field barrier or insulating region in the
`
`semiconductor, whereupon recombination can occur with majority carriers.
`
`But some materials resist injection of minority carriers even by this
`
`process, and then one must resort to the creation of the minority
`
`carriers within the material. There are two possible mechanisms for
`
`this:
`
`impact ionization and internal field emission.
`
`Impact ionization
`
`occurs when majority carriers are accelerated by the applied electric
`
`field to such high energies they can remove other carriers from a filled
`
`band by colliding with them.
`
`In this way minority carriers appear in
`
`the formerly filled band. Alternatively, it is possible to remove
`
`majority carriers from trap levels within the forbidden energy gap by
`
`this collision process, so that empty states are formed for recombi(cid:173)
`
`nation. With the internal field emission process, carriers are directly
`
`removed from a filled band or from filled localized levels by the action
`
`of the electric field (quantum mechanical tunneling). For materials
`
`where the band separation (or separation between a level and a band) is
`
`of sufficiently large energy so that visible light may be emitted
`
`during recombination, the calculated field strengths required for
`
`internal f ield emission appear to be prohibitively large.
`
`A device which also operates without minority carrier injection is
`
`the electroluminescent cell. Here a fine powder of a luminescent
`
`material is dispersed in an insulating plastic binder and placed between
`
`-6-
`
`TCL 1028, Page 30
`
`
`
`r
`
`two flat, transparent electrodes, f orming a capacitor. This assemblage
`
`is not conducting to a direct current, and so it is operated a-c. The
`
`subject of a-c electroluminescent cells has been the topic of a pro(cid:173)
`
`digious