Home

a study of efficiency droop of green light emitting

1. 10 N represents the group V elements such as P As N or Sb and H is hydrogen In the case of GaN trimethylgallium TMGa and ammonia NH3 react based on the following equation CH3 3Ga NH GaN 3 CH 2 5 The MOCVD growth process can be categorized into four processes gas input pyrolysis diffusion and surface reaction Then a by products formed during the epitaxial growth such as C2H6 are pumped away with carrier gases Figure 2 5 shows the schematic illustration of GaN epitaxial growth The MOCVD system used in our group is equipped with Close Coupled Showerhead CCS growth technology CCS technology can provide inherently growth uniformity due to precursor diffusion governed by mass transport from completely and uniformly intermixed gas phases Metalorganic group III precursors are introduced into the top plenum chamber and hydride precursors are introduced into the lower plenum chamber through a water cooled showerhead surface over the entire area of deposition The showerhead is constructed to enable precursors to be separated right up to the point where they are injected onto the substrates via high density of injection nozzles The complete intermixed uniform distribution of gas phases is created at approximately 5 mm out of total 11 mm spacing between the showerhead and a substrate The linear vertical temperature distribution in a growth chamber with CCS configuration can lead to high uniformity precursor decompositi
2. Precise lattice constant measurements provide information about lattice mismatch between the film and the substrate and therefore are indicative of strain and stress Superlattice measurements in multilayered heteroepitaxial structures which manifest as satellite peaks surrounding the main diffraction peak from the film are used to measure film thickness and quality But the most commonly scan technique used is the rocking curve measurement in which a scan at a fixed 20 angle is carried out In this configuration the detector is kept at a fixed angle relative to the primary beam 20 while the sample is rotated or rocked across the axis refer to Figure 4 8 Rocking curves are useful for determining the c lattice constant via the following equation nal In a perfect crystal with no strain or defects the w scan produces a 0 function at the Bragg angle But in the presence of imperfect plane spacing the Bragg peak is broadened The full width half maximum FWHM of this broadened peak is useful for comparing the relative quality of different samples Asymmetric rocking curves on planes can provide more information since the FWHM of these Bragg peaks are sensitive to threading dislocations 40 41 55 The following graph shows a high resolution XRD data The structure under measurement is a conventional LED structure composed of 5 InGaN 2 5 nm GaN 12 nm MQWs emitting at 500 nm GaN Si n type layer 3 um and two
3. 1 600E 06 y 2 495 54x 1 670 345 69 R 0 93 1 400E 06 1 200E 06 1 000E 06 8 000E 05 Integrated intensity a u 6 000E 05 Temperature K Raw IQE 48 49 IQE Linear 54 57 Figure 5 10 Example of Top a TD PL and bottom IGE result using copper tape The thermal contact is highly improved when using the grease as we can notice from the graph and IQE result where the difference between the raw data and the linearized one is only about 1 4 compared to around 6 for the previous result So far the maximum IQE measured is 56 sample 2 2506 3 490 nm peak wavelength with a MQW and the minimum is 26 sample 2 2461 3 525 nm peak wavelength with a MQW In conclusion the results confirm that when one approaches the green wavelengths an important decrease of the IQE is experienced from 50 for cyan LEDs around 490 nm to 35 for close green LEDs 525 nm due to the efficiency reduction experienced when approaching green wavelengths 71 6 QUICKTEST 2 0 The software named QuickTest 2 0 is a program based on the G language used by LabVIEW an integrated development environment developed by National Instruments The choice of using LabVIEW comes from the fact that the previous software was already a LabVIEW program and also because it is easier and faster to create a program with LabVIEW for instrument control rather than text based programs such as C C or Visual Basi
4. 17 I V Rozhansky D A Zakheim Analysis of Processes Limiting Quantum Efficiency of AlGaInN LEDs at High Pumping Phys Status Solidi A vol 204 no 1 pp 227 230 Jan 2007 18 I A Pope P M Smowton P Blood J D Thomson M J Kappers C J Humphreys Carrier Leakage in InGaN Quantum Well Light Emitting Diodes Emitting at 480 nm Appl Phys Lett vol 82 no 17 pp 2755 2757 Apr 2003 19 M H Kim M F Schubert Q Dai J K Kim E F Schubert J Piprek and Y Park Origin of Efficiency Droop in GaN Based Light Emitting Diodes Appl Phys Lett vol 91 pp 183507 1 3 Oct 2007 20 M F Schubert J Xu J K Kim E F Schubert M H Kim S Yoon S M Lee C Sone T Sakong Y Park Polarization Matched GaInN AlGaInN Multi Quantum Well Light Emitting Diodes With Reduced Efficiency Droop Appl Phys Lett vol 93 pp 041102 Jul 2008 21 Y C Shen G O Mueller S Watanabe N F Gardner A Munkholm and M R Krames Auger Recombination in InGaN Measured by Photoluminescence Appl Phys Lett vol 91 pp 141101 Oct 2007 22 A A Efremov N I Bochkareva R I Gorbunov D A Larinovich Yu T Rebane D V Tarkhin and Yu G Shreter Effect of The Joule Heating on The Quantum Efficiency and Choice of Thermal Conditions for High Power Blue InGaN GaN LEDs Semiconductors vol 40 no 5 pp 605 610 May 2006 23 S F Chichibu A Uedono
5. Appl Phys Lett vol 31 pp 201 203 1977 8 R D Dupuis P D Dapkus J N Holonyak E A Rezek and R Chin Room Temperature Laser Operation of Quantum Well Gag Al As GaAs Laser Diodes Grown by Metalorganic Chemical Vapor Deposition Appl Phys Lett vol 32 pp 295 297 1978 9 D Yoo Growth And Characterization of HI Nitrides Materials System For Photonic and Electronic Devices by Metalorganic Chemical Vapor Deposition Doctoral Dissertation Georgia Tech Aug 2007 10 A Khan K Balakrishnan T Katona Ultraviolet Light Emitting Diodes Based on Group Three Nitrides Nature Photonics vol 2 p 77 84 2008 11 P N junction Wikipedia Aug 2007 12 K Kumakura T Makimoto N Kobayashi Low Resistance Nonalloyed Ohmic Contact To P Type Gan Using Strained Ingan Contact Layer Appl Phys Lett vol 79 no 13 2001 88 13 United States Patent US6744799 freepatentonline com June 2004 14 C H Wang et al Appl Phys Lett vol 97 pp 261103 2010 15 B Monemar B E Sernelius Defect Related Issues in The Current Roll Off in InGaN Based Light Emitting Diodes Appl Phys Lett vol 91 pp 181103 Oct 2007 16 I V Rozhansky D A Zakheim Analysis of The Causes of The Decrease in The Electroluminescence Efficiency of AlGaInN Light Emitting Diode Heterostructures at High Pumping Density Semiconductors vol 40 no 7 pp 839 845 Jul 2006
6. C along with a reduction of growth pressure 75 Torr instead of 300 Torr leads to a stronger light emission The following figure displays the integrated EL intensity results at different injection currents LT EBL refers to the EBL grown at low temperature 780 C and HT EBL to the one at 840 C Plus a conventional structure has been grown without EBL for reference purpose w o EBL 0 10 20 30 40 50 60 70 80 90 Injection Current mA Figure 3 17 Integrated EL intensity function of injection current 36 The higher efficiency at higher injection currents in the case of LT EBL can be explained by a non uniform layer thickness when grown at low temperature enabling the holes to hop trough thinner regions of the EBL into the active region and then recombine radiatively with electrons But a higher peak wavelength shift compared to HT EBL has been observed probably due to hydrogen atoms from nitrogen precursor NH3 in MOCVD diffusing from the p layer to the QWs This diffusion can be responsible for a reduction of the spontaneous polarization induced internal electric fields in the active 37 region as the positive hydrogen ions cancel the fixed negative charges at the InGaN GaN interfaces This reduction of internal fields can contribute to the good performance observed with LT EBL as the non uniform layer thickness may facilitate the hydrogen ions penetration into the active region 3 9 4 Metallic Contact Device fabricatio
7. Nanotechnology Research Center Marcus Center 2 CRYSTAL GROWTH BASICS 2 1 HI Nitride Materials 2 1 1 Generalities direct and indirect bandgaps IlI Nitrides are a group of binary ternary and quaternary compound crystals formed by atoms of the group III in the Periodic Table 3 valence electrons and Nitrogen in group V 5 valence electrons Their major characteristics are large bandgap energies compared to GaAs for example high thermal stability and a direct bandgap which makes them ideal for optoelectronic applications In a direct bandgap semiconductor an electron in the conduction band can fall directly to an empty state in the valence band giving off the energy difference as a photon of light In an indirect bandgap semiconductor an electron in the conduction band must undergo a momentum change as well as changing its energy for example through a defect state before falling to the valence band The energy is usually given up as heat to the lattice rather than as an emitted photon Therefore a crystal with a direct bandgap can be used to create brighter LEDs while minimizing generation of heat The following figure illustrates the difference between direct and indirect bandgap Conductive Conductive Dey Direct radiative _ gt transition gt Indirect radiative m transition o T i ns l ui Momentum 5 Momentum Valence Valence band band Figure 2 1 left direct vs right indirect bandgap 2
8. 0 02 Joos 0 05 0 015 u_max pate p a Simulation speed Factor u Pump control u 4 suyas 0 01 J 0 02 effective after resstart 0 005 0 j 0 01012 umin A jo rho m3 J kla I Level h m A 0 53 esti AHh m Pi 0 90 Level Chart AL_h m iw 0 10 Ku J 5 D max m Levelh m J 1 13 0 94 0 8 0 7 A_init m 9 6 Jos z 0 5 0 54 0 42 0 3 0 24 O EIEEE IAE PCAC CAC RC CRC CECE CE CI G5 DHEA Aa DS OH 6 5 h_min m 0 1 A A Time s 7 o Ky t sec J 0 0005 47 1 g_out m3 s 0 0361377 Figure 6 1 Example of a front panel 45 The second is the block diagram where the source code is located It is the actual executable program The components of a block diagram are lower level VIs built in functions constants and program execution control structures Wires are drawn to connect the appropriate objects together to indicate the flow of data between them Front panel objects have corresponding terminals on the block diagram so data can pass from the user to the program and back to the user 73 Pulldown Menus Panel Palette gt Simple Adding Machine vi Diagram Control Terminais Indicator Terminals Loop Node Boolean Structure Wires Switch Figure 6 2 Example of a block diagram 46 6 2 Motivation The motivation was to replace the previous software that was not optimized for Electroluminescence application I
9. In our setup 4 different instruments are connected to the computer for different measurement purposes EL TLM oscilloscope and capacitance measurement station Each instrument has a unique assigned address The dentification string indicator permits to verify that the communication has actually been established T1 The program sends an identification request IDN then the source meter replies back with the following identification string KEITHLEY INSTRUMENTS INC MODEL 2430 6 3 2 Sample information and Spectrometer parameters The User s Initial Substrate ID and Epi Run ID are the identification information about the user and the sample under measurement They are saved in the spectrum and data files headers Concerning HR2000 S N it is the serial number of the Ocean Optics spectrometer This information indicates that the spectrometer has been recognized by the computer The Integration Period specifies the integration time in ms of the spectrum acquisition by the spectrometer It is adjusted so that the anticipated greatest amount of light causes a signal of about 3500 counts The Average specifies the number of discrete spectral acquisitions that the device driver accumulates before the software receives a spectrum The higher the value the better the signal to noise ratio S N but the longer the acquisition The S N will improve by the square root of the number of scans averaged The Boxcar Smooth
10. Rinke K T Delaney C G Van de Walle Indirect Auger Recombination as a Cause of Efficiency Droop in Nitride Light Emitting Diode Appl Phys Lett vol 98 pp 161107 2011 34 E Kioupakis P Rinke A Schleife F Bechstedt and C G Van de Walle Phys Rev B vol 81 pp 241201 2010 35 T Li A M Fischer Q Y Wei F A Ponce T Detchprohm C Wetzel Carrier Localization and Nonradiative Recombination in Yellow Emitting Ingan Quantum Wells Appl Phys Lett vol 96 pp 031906 2010 36 A M Fischer K W Sun E Juday F A Ponce J H Ryou H J Kim S Choi R D Dupuis Effect of Growth Temperature on the Electron Blocking Performance of InAIN Layers in Green Emitting Diodes Appl Phys Express vol 3 pp 031003 2010 37 Atomic Force Microscopy Wikipedia 2009 38 NPTEL online classes http nptel 1itm ac in courses Webcourse contents IIT Delhi Semiconductor 20Devices LMB2A 2d htm 39 University of Santa Barbara Introduction to X Ray Diffraction http www mrl ucsb edu mrl centralfacilities xray xray basics index html 90 40 B Heying X H Wu S Keller Y Li D Kapolnek B P Keller S P DenBaars and J S Speck Role of Threading Dislocation Structure on The X Ray Diffraction Peak Widths in Epitaxial GaN Films Appl Phys Lett vol 68 pp 643 645 1996 41 C Ryang Wie High Resolution X Ray Diffraction Characterization of Semiconductor Structures Ma
11. The bandgap energy JE is directly linked to the wavelength of emission A through the following equation 2 1 ip A Where A is the Planck constant and c is the speed of the light in free space The HI Nitride alloys that are of interest in this thesis are GaN InGaN AlGaN and InAIN A GaN crystal has a larger bandgap energy 3 4 eV compared to GaAs 1 43 ev which corresponds to UV wavelengths However when indium In or aluminum Al is added to GaN the bandgap energy can be engineered to allow photon emission in the visible spectrum depending on the composition of each additional element in the GaN based alloy In or Al Refer to the following figure for the spectrum range that InGaN AlGaN and AlInN cover lil V nitrides T 300K Alw Ing oN UV Violet Bandgap energy E eV Wavelength A nm IngyGacxN 3 0 z i HF 3 3 3 4 Lattice constant ag A tad Fi tad co td Figure 2 2 Spectrum of emission in function of In Al and Ga concentrations in lIl Nitrides alloys 3 The bandgap energy E in eV can be determined based on the composition x in percentage of each element in the alloys using the Vegard s law Eg n x Ga 1 xN x Eg 1 x Eg can b x 1 x 2 2 Ain x Ga 1 x N X X ammn 1 x agan b x 1 x 2 3 The variable a in A is the lattice constant of the crystal The parameter b in eV is called bowing parameter In practic
12. _ oOo 350 400 450 500 550 600 650 700 Peak Wavelength nm Figure 1 1 External Quantum Efficiency for the visible spectrum 1 Therefore the challenge we are facing today is to fabricate green LEDs that can emit light as bright as blue and red ones to obtain white color based on the RGB Red Green Blue combination The major difference between red and blue green LEDs is that the former is made of InAlGaP grown on GaAs and the latter of InGaN grown on GaN 1 2 Overview of AMDG AMDG stands for Advanced Materials and Devices Group and is under the direction of Dr Russell Dupuis and the supervision of Dr Jae Hyun Ryou So far the group possesses two cleanrooms The first one is composed of three Metalorganic Chemical Vapor Deposition MOCVD reactors one dedicated for GaAs related material growth and the other two for GaN related materials The second cleanroom is dedicated to device testing and material characterization A list of equipment will be described in the Chapter 4 An optics lab is also used for photoluminescence measurements The current main focus of the group is to grow III Nitride devices such as blue green LEDs UV Laser Diode LD structures along with HFETs and HBTs The crystal growth is performed in the group s cleanroom while device measurement and characterization mostly in our device testing lab and the device fabrication process is performed in the Microelectronics Research Center MiRC as well as the
13. cell 5 The major issue to overcome with the group III Nitrides is lack of a large area high quality lattice matched substrate Indeed it is of prime importance to have an excellent crystal quality in order to increase device lifetime and efficiency by limiting or even avoiding dislocations due to lattice mismatch between the substrate and layers above Obviously a GaN substrate allows a better lattice matching for visible LEDs which reduces the polarization effect due to the strain free interface However the fabrication of free standing GaN substrate wafers is complex typically vapor phase transport or epitaxial type growth is used and only a small number of companies produce them leading to a cost prohibitive price several thousand dollars for a 2 wafer for the mass production of LED devices Among other substrates the most commonly and widely used is 0001 sapphire due to its hexagonal symmetry thermal stability availability and low cost However the major drawback is a large lattice mismatch with 0001 GaN around 13 which results in crystal strain and defect formation in epitaxially grown layers 2 1 3 Doping of GaN Intrinsically GaN is an n type semiconductor However because of a relatively low background carrier concentration 10 to 10 cm GaN is doped in order to increase the free carrier concentration Silicon and germanium are the two main n type dopants but silicon is more generally used b
14. different p type layers p GaN for reference and p Ing 92Gao 9gN LED w p GaN LED w plInGaN X ray intensity arb unit 6000 4000 2000 a 2000 4000 6000 Angle separation arc seconds Figure 4 10 Data result from XRD measurement Conventional LED structure with bottom p GaN and top p InGaN AMDG Internal Document 56 5 TEMPERATURE DEPENDENT PHOTOLUMUNESCENCE The following sections Chapters 5 and 6 describe the actual work I have done while in the AMDG They are ordered in a chronological order beginning with a Photoluminescence measurement station I have setup to provide temperature dependent PL data in order to primarily determine the internal quantum efficiency IQE of our LED structures The IQE is the ratio of the photons emitted from the active region to the number of electrons photogenerated into the LED structure The assumption is at very low temperature every photogenerated electron gives birth to a photon which leads to 100 IQE But as the temperature increases nonradiative recombination occurs in the active region due to defect sites that are thermally activated as well as other effects we have discussed related to the efficiency droop Therefore the IQE gives us a good way to compare the quality of the MQW structure of different samples An IQE of 100 means that the crystal is perfect no defects and only radiative recombination occurs But in practice the IQE determined at 300K in
15. dissipated by the excitation of a free electron high into the conduction band or by a hole deeply excited into the valence band The highly excited carriers will subsequently lose energy by multiple phonon emission until they are close to band edge The Auger recombination rate can be expressed as 30 Unuger Cn x An x no Any 3 1 where C represents the Auger recombination coefficient for electrons n the equilibrium and An the excess carrier concentrations An increased current in an LED leads to an increased carrier concentration in the active region sO one expects that Auger recombination will become an important loss factor at some stage due to its dependence to the third power of the carrier density 27 At high current injection densities the Auger recombination rate can be expressed by the following equation For An gt gt no Usuger Cn X An 3 2 As it is dependent to the third power of the carrier density this can lead to a possible main source for the efficiency droop in our LEDs because the multiple quantum well structures that we mostly use confine the carriers in a small volume which results in higher carrier densities compared to conventional double heterostructure active regions However Auger recombination is greater for relatively small bandgap materials compared to large bandgap semiconductors such as GaN In fact the Auger recombination carrier lifetime is given by a modified expression of Bea
16. is to determine the internal quantum efficiency by comparing the integrated intensity of emission at different temperatures 10 K to 310 K with a 20 K step thanks to a temperature controlled cryostat system The following diagram shows the setup of the TD PL system 58 Mirror I Monochromator CCD Detector Data Acquisition Card ADC Cold Finger LED Sample Figure 5 1 Schematic diagram of the TD PL setup 5 2 Optical pump The choice of excitation is critical in any PL measurement The excitation energy and intensity will have profound effects on the PL signal Because the absorption of most materials depends on energy the penetration depth of the incident light will depend on the excitation wavelength Hence different excitation energies probe different regions of the sample The excitation energy also selects the initial excited state in the experiment Because lasers are monochromatic intense and readily focused they are the instruments of choice for PL excitation For our applications the excitation power is not critical as the active region of the samples under measurement is the first region that the laser strikes Indeed many of the samples used for PL measurements are grown without a p type layer 59 which enables the laser to excite all of the QWs in the active region This has been proved by using a Triple Wavelength MQW where the emission of three peak wavelengths corresponding
17. our LED structures is about 35 for peak wavelengths around 525 nm and roughly 50 for 490 nm The measurement has been carried out for mostly MQW active regions and no p type layer 57 5 1 Preliminary Temperature Dependent Photoluminescence TD PL is a technique used to acquire the spectrum from samples when optically excited at different temperatures As introduced in section 4 2 Photoluminescence PL is the spontaneous emission of light from a material under optical excitation PL analysis is nondestructive because the technique requires very little sample manipulation or environmental control Because the sample is excited optically electrical contacts and junctions are unnecessary and high resistivity materials pose no practical difficulty When light of sufficient energy is incident on a material photons are absorbed and electronic excitations are created Eventually these excitations relax and the electrons return to the ground state or valence band If radiative relaxation occurs the emitted light 1s called PL The PL spectrum provides the transition energies which can be used to determine electronic energy levels The PL intensity gives a measure of the relative rates of radiative and nonradiative recombination Compared with other optical methods of characterization like reflection and absorption PL is less stringent about beam alignment surface flatness and sample thickness For our application the first objective
18. quantum barrier one quantum well is equal to approximately 15 nm Depending on the indium composition of the alloy the bandgap energy can be engineered to target the desired wavelength of emission MQW structures enhance the confinement of carriers in multiple thin regions where injected carriers can recombine at a higher rate than for a SQW because of higher carrier densities inside wells wells in MQW structures are thinner than for SQWs 18 j ns 1 a saa 3 i n i si aS k n pi ore EBEEEL FH Figure 3 3 Illustration of a MQW structure 13 But increasing the carrier density enhances nonradiative recombination such as Auger recombination which leads to a degradation of performance Furthermore because of a difference of effective masses between holes and electrons the recombination occurs mainly close to the p type layer instead of having a uniform distribution over the SQW or trapped in all QWs in the MQW structure 3 6 Electron Blocking Layer Because the effective mass of electrons is seven times lighter than holes electrons have a tendency to propagate faster Therefore some electrons recombine in the p type layer or reach the metallic contact leading to a leakage current This undesirable current flow is referred as a current overflow or carrier spillover see section 3 8 2 One way to tackle this issue is to incorporate an electron blocking layer EBL that reduces the flow of electrons throug
19. 50 100 150 200 Current Density A em Figure 3 5 Output power function of current density for conventional and GEBL LEDs adapted from 14 3 7 Device Fabrication The device fabrication is done after the growth of the entire LED structure It is composed of four steps Mesa etching N contact deposition P spreading metal deposition and P bonding pad deposition Mesa etching 1 Mesa etching is a technique used to pattern the LED wafer structure in order to create regions for placing the ohmic contacts to the n type and p type layers The first step is a photolithography spreading a photoresist PR coating over the sample surface alignment of the mask and PR exposure to UV light PR development deposition of S102 and finally PR removing Then dry etching is done by Inductively Coupled Plasma ICP technique in our cleanroom 21 d Buffer Layer undoped GaN Buffer Layer undoped GaN Sapphire Substrate Sapphire Substrate Figure 3 6 Illustration of the mesa etching not in scale P type layer N type layer Figure 3 7 Image of the sample surface after mesa etching Courtesy of AMDG N contact metal deposition 2 The second step is to deposit a metallic contact on the n type layer using an e beam evaporator An E beam evaporator is a physical vapor deposition system that heats up a metal in the solid state by bombarding the target with a highly energetic electron beam produced by a charged tungsten filame
20. 530 nm 48 The data for blue LEDs has been published in 2007 and the green one in 2009 Finally a new Lab VIEW program has been developed for EL measurement station The motivation was to automate the entire station in order to drastically decrease measurement time and to obtain more accurate results thanks to a better repeatability 87 REFERENCES 1 M R Krames O B Shchekin R Mueller Mach G Mueller L Zhou G Harbers and M G Craford Status and Future of High Power Light Emitting Diodes for Solid State Lighting Journal Display Technology vol 3 pp 160 175 2007 2 J Hecht Photonic Frontiers Silicon Photonics Closing in on Silicon Lasers Laser Focus World Feb 2006 3 C L Progl High Resolution Electron Beam Testing of Gallium Nitride Based Light Emitting Diodes Doctoral Dissertation North Carolina State University 2008 4 Application Oriented Quantum Theory for Infrared Nitride Lasers Semiconductor Today February 2010 5 Z Lochner Green Light Emitting Diodes And Laser Diodes Grown By Metalorganic Chemical Vapor Deposition Master Thesis Georgia Tech May 2010 6 H M Manasevit Single Crystal Gallium Arsenide on Insulating Substrates Appl Phys Lett vol 12 pp 156 159 1968 7 R D Dupuis P D Dapkus R D Yingling and L A Moudy High Efficiency GaAlAs GaAs Heterostructure Solar Cells Grown by Metalorganic Chemical Vapor Deposition
21. A STUDY OF EFFICIENCY DROOP OF GREEN LIGHT EMITTING DIODES GROWN BY METALORGANIC CHEMICAL VAPOR DEPOSITION A Master Thesis Presented to The Academic Faculty By Nordine Sebkhi In Partial Fulfillment Of the Requirement for the Degree Master of Science in the School of Electrical and Computer Engineering Georgia Institute of Technology December 201 1 A STUDY OF EFFICIENCY DROOP OF GREEN LIGHT EMITTING DIODES GROWN BY METALORGANIC CHEMICAL VAPOR DEPOSITION Approved by Dr Russell Dupuis Advisor School of Electrical and Computer Engineering Georgia Institute of Technology Dr Shyh Chiang Shen School of Electrical and Computer Engineering Georgia Institute of Technology Dr Paul Douglas Yoder School of Electrical and Computer Engineering Georgia Institute of Technology Date Approved ACKNOWLEDGEMENTS I would like to begin by thanking everyone who made my work possible and enriching First of all I am grateful to my advisor Dr Russell Dupuis for giving me the priceless opportunity to work at the Advanced Materials and Devices Group In addition to his support and help he has provided me with his expertise on MOCVD and LEDs to thrive and gain an invaluable experience Also I would like to thank Dr Jae Hyun Ryou for the opportunity to have worked on interesting and meaningful projects and for his support Then I would like to express my thanks to my colleagues Jeomoh Kim Mi Hee Ji and Zachary Lochner wi
22. AFM is a very high resolution type of scanning probe microscopy SPM with demonstrated resolution on the order of fractions of a nanometer The first atomic force microscope was invented in 1986 by Binnig Quate and Gerber and the first commercially available one was introduced in 1989 The AFM is used for imaging measuring and manipulating matter at the nanoscale For our applications the AFM is operated as a sample surface imaging The information is gathered by scanning the surface with a mechanical probe Piezoelectric elements that 43 facilitate tiny but accurate and precise movements on command enable the very precise scanning In general an AFM system is composed of five major components a cantilever probe tip X Y stage laser Position Sensitive Photo Diode PSPD and feedback mechanism Detector and Solid State Laser Diode Feedback Electronics Photodiode danois Stay Cantilever amp Tip PZT Scanner Soe Cantilever and Tip Figure 4 4 Left Overview of AFM system Right Position detection mechanism 37 When the sharp tip made from silicon by micromachining is scanned over a surface with feedback mechanism that enables the piezoelectric Z scanner to maintain the tip at a constant force to obtain height information the flexible cantilever under which the tip is attached moves up and down with the contour of the surface A laser beam focused on the back of reflective cantilever is deflecte
23. Al to the AlGaN to increase the bandgap also increases these polarization field effects It 1s also difficult to grow good quality high Al content AlGaN layers The optimum growth temperature for AlGaN can also be quite high producing significant damage in the underlying layers particularly for the high In content of InGaN in the active region InAIN can be lattice matched to GaN and InGaN reducing the piezoelectric effects In addition to having a larger energy bandgap InAIN also has a large conduction band offset relative to GaN presenting a large barrier to electrons entering the p type region and relatively small valence band offset so that holes can get into the active region Growth of InAIN also takes place at a lower temperature than AlGaN leading to less damage to the active region The following figure shows the improvement of performance by comparing different structures InAIN composed of 18 of indium and grown at 845 C AlGaN composed of 20 of aluminum and grown at 930 C and non EBL structure with 20nm of p GaN LED with inAIN Eel a Ps 4 s A reee a P A a7 LED without EBL a o 4 g Light output a u 0 50 100 150 200 250 300 350 400 450 Injection current mA Figure 3 16 LED performances without EBL with AlGaN and InAIN EBL AMDG Internal Document 36 Furthermore a study of InAIN EBL epitaxial growth in our group 36 shows that a reduction of the growth temperature from 840 C to 780
24. C or pulse mode The EL spectrum is acquired by a spectrometer Ocean Optics HR2000 connected by USB to a computer Below is a diagram of how light propagates through the optical bench of the HR2000 Spectrometer Figure 4 2 left Schematic and right picture of the HR2000 Spectrometer HR2000 User s Manual 40 The HR2000 Component Table on the following page explains the function of each numbered component in the HR2000 Spectrometer schematic SMA Connector Secures the input fiber to the spectrometer 5 Slit The size of the aperture regulates the amount of light that enters 1 the optical bench and controls spectral resolution Restricts optical radiation to pre determined wavelength regions 3 Filter Both bandpass and long pass filters are available to restrict radiation to certain wavelength regions Focuses light entering the optical bench towards the grating of the Collimating Mirror E E P E E spectrometer 5 Gratin Diffracts light from the Collimating Mirror and directs the j diffracted light onto the Focusing Mirror 6 Focusing Mirror Focuses the light onto the L2 Detector Collection Lens Focuses light from a tall slit onto the shorter CCD Detector L2 Detector Collection Lens elements Collects the light received from the L2 Detector Collection Lens CCD Detector and converts the optical signal to a digital signal Each pixel on the CCD Detector responds to the wavelength of light that strikes it creat
25. Figure 3 10 Left Schematic and Right image after p bonding pad deposition 24 Figure 3 11 Illustration of the Efficiency Droop ccecccccccccccesseeeseeceeeeeeeaaeeeseeceeeeeeeaas 25 Figure 3 12 Illustration of a direct and b indirect Auger recombination 5 29 Figure 3 13 Spatial separation of electron and hole wave functions in QCSE 32 Figure 3 14 T W MQWs EL spectrum with left p GaN and right p InGaN 34 Figure 3 15 TW MQWs EL spectrum with In composition of 1 5 2 and 3 34 Figure 3 16 LED performances without EBL with AlGaN and InAIN EBL 36 Figure 3 17 Integrated EL intensity function of injection current ccceeeeeeeeeeeeees 37 Figure 4 1 Illustration of radiative recombination process ccccccesssssseeeeeeeeeeeeeeeeeeeees 39 Figure 4 2 left Schematic and right picture of the HR2000 Spectrometer 40 Figure 4 3 Illustration of photon generation by Photoluminescence ccccceeeeeeees 42 Figure 4 4 Overview of AFM system and position detection mechanism 00 44 Figure 4 5 AFM Image of a GaN surface with scale scan parameters and data result 47 Fiure 40 Example On TEM TeS serai iscuteietiaiudealchaeceressacesmsaumeiesaatvdeetenene 48 Figure 4 7 Illustration of Hall effect in a p type baf nenssssseseenesssssssseeerrsssssssseersssssss 49 Fiure A S Schema
26. K carriers are able to overcome potential fluctuations and relax into lower potential minima where the indium composition is higher leading to a red shifted spectrum of emission Then if the temperature keeps increasing the lower bandgap energy states will be populated so a delocalization out of potential minima occurs and results in blue shift This process is referred as S shape temperature dependence of the peak intensity 43 66 Conduction Band Temperature lt 70 K amp Electron hole pair generation Radiative recombination ye Ce iene Valence Band Temperature 70 150 K Temperature gt 150 K Figure 5 6 Illustration of indium localization effect 67 Localized potential minima prevent carriers from reaching dislocations and instead recombine radiatively By this means indium localization enhances photon emission and so improves brightness But the growth of high quality InGaN becomes increasingly more difficult as the In composition increases In fact lattice mismatch resulting in increasing polarization effect and poor crystal quality become predominant and so degrade performances Studies show that indium localization improves brightness for blue LEDs but degrades efficiency when approaching green wavelengths 44 35 Another data result of interest 1s the linewidth or Full Width Half Maximum FWHM This is an assessment of the crystal purity as impurities create defect sites resul
27. T Onuma S Nakamura S Yamaguchi S Kamiyama H Amano I Akasaki Origin of Defect Insensitive Emission Probability in In Containing Al In Ga N Alloy Semiconductors Nat Mater vol 5 no 10 pp 810 816 Oct 2006 24 S F Chichibu T Sota K Wada and S Nakamura Exciton Localization in InGaN Quantum Well Devices J Vac Sci Technol B vol 16 no 4 pp 2204 2214 Aug 1998 89 25 A R Vasconcellos R Luzzi C G Rodrigues and V N Freire Hot Phonon Bottleneck in The Photoinjected Plasma in GaN Appl Phys Lett vol 82 no 15 pp 2455 2457 Apr 2003 26 J H Ryou P D Yoder J Liu Z Lochner S Choi H J Kim R D Dupuis Control of Quantum Confined Stark Effect in InGaN Based Quantum Wells IEEE Jour of Sel Topics in Quant Elec vol 15 no 4 Aug 2009 27 Cooke Solutions Don t Solve Droop Controversy Semiconductor Today Compounds and Advanced Silicon vol 3 issue 9 Nov 2008 28 Shen et al Appl Phys Lett vol 91 2007 29 Niet al Appl Phys Lett vol 93 2008 30 U Ozgur H Liu X Li X Ni H Morkoc GaN Based Light Emitting Diode Efficiency at High Injection Levels Proceedings of the IEEE 2009 31 A R Beattie P T Landsberg Auger Effect in Semiconductor Proc R Soc Lond A vol 249 1959 32 J Hader J V Moloney B Pasenow et al Appl Phys Lett vol 92 261103 2008 33 E Kioupakis P
28. a technique used to determine the contact resistance between a metal and a semiconductor The technique involves making a series of metal semiconductor contacts separated by various distances Probes are applied to pairs of contacts and the resistance between them is measured by applying a voltage across the contacts and measuring the resulting current The current flows from the first 47 probe into the metal contact across the metal semiconductor junction through the sheet of semiconductor across the metal semiconductor junction again into the second contact and from there into the second probe and into the external circuit to be measured by a current meter The resistance measured is a sum of the contact resistance of the first contact the contact resistance of the second contact and the sheet resistance of the semiconductor in between the contacts If several such measurements are made between pairs of contacts that are separated by different distances a plot of resistance versus contact separation can be obtained The contact separation is expressed in terms of L gap between 2 contacts W is the width of the area between the contacts Ideally the plot should be linear and the result of the multiplication between the slope of the line and the width W is the sheet resistance The intercept of the line with the y axis is two times the contact resistance Thus the sheet and contact resistance can be determined from this technique
29. al layers suffered from impurity related defects In 1977 Dupuis et al 7 8 have demonstrated the first practical devices with AlGaAs GaAs solar cells and quantum well injection lasers The fundamental principle of the MOCVD is to introduce two or more materials in a gaseous form into a reaction chamber where they chemically react with one another to form a new material deposited on the wafer surface The MOCVD growth technique involves a sequence of chemical reactions among different precursors A metalorganic precursor exists in either liquid or solid form and 1s stored in an all welded stainless steel container commonly referred to as bubbler A carrier gas passes or bubbles through the precursor container and carries metalorganic precursor molecules into an epitaxial growth chamber thus precursors must exhibit an appropriate volatility Also precursors should have the proper reactivity to thermally decompose in an epitaxial growth chamber Trimethylgallium TMGa triethylgalltum TEGa trimethylaluminum TMAI and trimethylindium TMIn are commonly used as group III precursors Concerning the group V sources Nitrides the use of ammonia NH3 is the most common The general reaction to form group IIJ Nitrides is described below RM NH e MN 3 RH 2 4 In this equation R is an alkyl group for instance methyl CH3 or ethyl C2H5 and M represents the group III metal such as gallium Ga aluminum Al or indium In
30. along with the 86 device fabrication process Hole transport is currently under intense investigation as it may be one of the major reasons for efficiency droop Therefore different samples are growth with p GaN and p InGaN layers in order to study the difference in hole transport Research on the design of the Electron Blocking Layer using InAIN instead of AlGaN resulted in the observation of increase of light intensity by comparing Electroluminescence measurement results of LED structures with EBL made of InAIN versus AlGaN and no EBL Device fabrication is of prime importance too on the effort of improving light emission by reducing the metal contact resistance on the p type layer side leading to a better hole injection This has been done by decreasing the Nickel layer thickness from previously 5 nm to 2 nm Changing the annealing temperature from currently 500 C to 550 C may permit to further decrease contact resistance The setup of a Temperature Dependent Photoluminescence test station enables us to gather important data about the quality of the active region by assessing the internal quantum efficiency The first set of data shows an IQE of 50 for peak wavelengths of around 490 nm and roughly 35 for 525 nm confirming the degradation of LED performance when approaching green wavelengths In the literature the highest IQEs published for blue LEDs 450 nm is around 70 47 around 50 for 490 nm 48 and 40 for green wavelengths
31. applying a non linear voltage to the piezo electrodes to cause linear scanner movement and calibrating the scanner accordingly AFM has three modes of operation contact mode non contact mode and tapping mode In contact mode the force between the tip and the surface is kept constant during scanning by maintaining a constant deflection thanks to the feedback mechanism The tip and the surface are actually in contact which means the necessity of having a hard sample surface in order to not deteriorate the sample In non contact mode the tip of the 45 cantilever does not contact the sample surface The cantilever is instead oscillated at a frequency slightly above its resonant frequency The van der Waals forces which are strongest from nm to 10 nm above the surface or any other long range force which extends above the surface acts to decrease the resonance frequency of the cantilever This decrease in resonant frequency combined with the feedback loop system maintains a constant oscillation amplitude or frequency by adjusting the average tip to sample distance Measuring the tip to sample distance at each x y data point allows the scanning software to construct a topographic image of the sample surface In tapping mode the idea is to keep the probe tip close enough to the sample for short range forces to become detectable while preventing the tip from sticking to the surface The cantilever is in tapping mode driven to oscillate a
32. aver sacnieteh sateen aiid A 16 S ed B14 0 oe Be fc emma E A rere ea ney Sere Re a Teen TS etre el eee nT ene Tenn Serer eee mere Tee 17 Fe hy ANC AYA oil nlc 24 0 Sane ee Eee et eee SOE RE Cot Pent enn nna ee STE nT tor emt ene a ona en yee mnt rie 18 9 6 Tlectron BIOCKINe Ia Ver sia ecsaiedensasatacnecanzactaagideatabunteauatataenestazactaauisentebaemealaoweieis 19 Diol Device Fabric anO cies esos a a oe teach sndecds cadcdebeaeaess 21 OO PCI EN DOOP sacespicaiee sca aeaawuciats E NE 25 3 8 1 Auger Recombination ccccccccccccscccceeeessssssseeeececceccceeesseaaceeessseseeeeeeeeess 21 ISe Care OVET ON oreinen a 30 3S r POVANIZ AOU CLIC aea thir east ae rn hatha Sahat Aue sti 31 39 aprove mento LEDS nyaa tein aat area scen teehee 32 D2 Only GIO Le Tapo eenaa O aes 32 5 002 VE Gavan TMG PrecursOrs cipere r na a 35 3 0 3 Electron Blockins Layer Study seesnoiuie ree E satis eetiuxnsaeveeveacs 35 1V 38 4 Metalice Contat ca vcsscees ce nckaedts sein eebentodsated sdoucean tie ee doiaseeeewidtuved NaS 38 CHAPTER 4 EQUIPMENT FOR MEASUREMENT PURPOSE ccccccccccccccccceceeeeeeeeeeeees 39 AW TLCCIFOMMMINCSCOMCE sucen nani edie ai oheldaaunindssdatiwncenan Melee nioeeuseaeedalsaaded 39 Ac PUOUOIUMIME SCONCE 5s teneattatcerae atta erin a ene ian cat aan es ease ean ea 4 A Atomic FOCE MICTOSCOPY arasina aa ee aha 43 E silts WA sag sess ncaa cuetacas E guntast AE EE E E E E sotceananauses 47 AS Hall Effect M
33. c Besides as the communication between the computer and the current source is using GPIB General Purpose Interface Bus Lab VIEW has built in functions for GPIB communication that makes the programming easier 6 1 LabVIEW LabVIEW stands for LABoratory Virtual Instrument Engineering Workbench It is a programming environment in which programs are created with graphics It is designed for scientists and engineers who need to program as part of their jobs The graphical programming language called G enables the users to create programs in a pictorial form called a block diagram eliminating a lot of the syntactical details LabVIEW uses terminology icons and ideas familiar to scientists and engineers It relies on graphical symbols rather than textual language to describe programming actions The core of the G language is based on the principle of dataflow in which functions execute only after receiving the necessary data governs execution in a straightforward manner 12 LabVIEW programs are called Virtual Instruments VIs because their appearance and operation imitate actual instruments VIs has primarily two main parts The front panel is the interactive user interface of a VI so named because it simulates the front panel of a physical instrument It can contain knobs push buttons graphs and many other controls which are user inputs and indicators which are program outputs Tank Simulator Time_step s
34. ces 52 which are thousands to millions of times more intense than laboratory X ray tubes have become indispensable tools for a wide range of structural investigations X rays primarily interact with electrons in atoms When X ray photons collide with electrons some photons from the incident beam will be deflected away from the direction where they originally travel If the wavelength of these scattered X rays did not change meaning that X ray photons did not lose any energy the process is called elastic scattering Thompson Scattering in that only momentum has been transferred in the scattering process These are the X rays that we measure in diffraction experiments as the scattered X rays carry information about the electron distribution in materials On the other hand in the inelastic scattering process Compton Scattering X rays transfer some of their energy to the electrons and the scattered X rays will have different wavelength than the incident X rays Diffracted waves from different atoms can interfere with each other and the resultant intensity distribution is strongly modulated by this interaction If the atoms are arranged in a periodic fashion as in crystals the diffracted waves will consist of sharp interference maxima peaks with the same symmetry as in the distribution of atoms Measuring the diffraction pattern therefore allows us to deduce the distribution of atoms in a material The peaks in an X ray diffractio
35. ckground lighting in handheld electronic devices Their high switching rates are also useful in advanced communications technology as optical emitters for emerging fiber optic networks Infrared LEDs are also used in the remote control units of many commercial products Currently LEDs are being developed for general lighting applications where their particular characteristics such as efficiency longevity durability compactness and cool operation are needed The worldwide lighting market will grow to approximately 159 billion in 2020 with 80 of that total from general lighting according to a report from McKinsey amp Company LEDs magazine Lighting market report predicts strong growth for LED lighting September 2011 However to be able to lead the technological revolution in the general lighting market LEDs need to cover the full visible spectrum in order to emit true white color So far red and blue LEDs are well mastered but when one approaches green wavelengths their efficiency decreases This is the so called green gap due to the internal quantum efficiency reduction as the LED active region design incorporates more In into the active region Efficiency droop affects virtually all II N LEDs but is much more pronounced as the wavelength gets longer However it is still a significant effect even for blue LEDs 60 Ga In N mens Al In Ga P No wo ta g1 ea O External Quantum Efficiency
36. cture have been fabricated but show different EL results The question is whether this difference is due to 34 growth condition change or fabrication process variation as the EL acquisitions have been done after device fabrication Therefore the mechanisms responsible for this unexpected behavior are not understood and are currently under intense investigation 3 9 2 TEGa and TMGa Precursors TEGa and TMGa are typical precursors for Ga in MOCVD No difference is experienced on the emitted intensity between using TEGa and TMGa for MQWs having a peak wavelength at around 500 nm However according to QuickTest EL results there was an enhancement of intensity by using TEGa for MQWSs emitting at 470 nm The improvement of intensity is about 19 6 when the injected current is 80 mA and 25 at 20 mA But this observation needs to be confirmed after device fabrication because metallic contact used before device fabrication are not of a good quality possibly leading to inaccurate results 3 9 3 Electron Blocking Layer Study This study is about using InAIN instead of AlGaN as an EBL One problem with AlGaN EBL is that strain effects from lattice mismatch between the EBL and GaN from the active region can create strain based piezoelectric fields in the active region Further problems arise from the different spontaneous polarization fields These fields act in a 35 sense that reduces the barrier s effectiveness Adding more
37. d for our application the detector is cooled using liquid nitrogen through a 1 liter reservoir capable of reaching 130 C and lasts about 24 hours The voltage is then digitalized through an Analog to Digital Converter ADC before being formatting into a data acquisition card connecting to a computer The spectrum scanned during this process is from 200 nm to 1200 nm but the actual data used by the group is from 350 nm to 650 nm 5 5 TD PL data result The following results show the measurement data of the LED sample no 2 2506 3 with a conventional MQW structure composed of 5 InGaN quantum wells 63 60000 40000 20000 Intensity Counts 450 458 467 475 483 492 500 508 517 525 533 542 550 Wavelength nm Intensity vs Temperature y 131 45x 64419 R2 0 9761 Figure 5 4 Top TD PL and bottom intensity function of temperature of the sample This previous figure shows an overlaid graph of PL measurement at different temperatures At first one can notice that a factor of 3 exists between the peak intensities at the minimum 10 75 K and maximum 310 K temperatures which prove a strong temperature dependence of nonradiative recombination 64 Then a red shift trend is experienced when increasing temperature This is explained from Varshni formula 5 1 Where a and p are Varshni fitting parameters The energy bandgap reduces with increasing temperature explai
38. d toward the PSPD The photodetectors measure the deflected light intensities and convert the cantilever motion into voltage and this is converted into height information using a differential amplifier 44 The use of a feedback mechanism is due to the fact that if the tip was scanned at a constant height a risk would exist that the tip collides with the surface causing damage Hence a feedback mechanism is employed to adjust the tip to sample distance to maintain a constant force between the tip and the sample The sample is mounted on a piezoelectric stage that can move the sample in the z direction for maintaining a constant force and the x and y directions for scanning the sample The tip is mounted on a vertical piezoelectric scanner while the sample is being scanned in X and Y using another piezoelectric block The resulting map of the area z f x y represents the topography of the sample Scanners are characterized by their sensitivity which is the ratio of piezo movement to piezo voltage 1 e by how much the piezo material extends or contracts per applied volt Because of differences in material or size the sensitivity varies from scanner to scanner Sensitivity varies non linearly with respect to scan size Piezo scanners exhibit more sensitivity at the end than at the beginning of a scan This causes the forward and reverse scans to behave differently and display hysteresis between the two scan directions This can be corrected by
39. e the curves are not linear like in the Figure 2 2 but are bowed The bowing parameter is determined experimentally according to the crystal quality and the alloy composition The following figure shows the impact of different values of the bowing parameter In general the bowing parameter is set at around 1 eV AIGaN b 1 DeV InGaN be 1 2eV IAAIN b 2 5eV InAIN b 3 0eV IRAIN b 3 5eV InAIN b 4 0e InAIN b 4 5eV Bandgap eV Lattice constant e A Figure 2 3 NALIN and InGaN Bandgaps with different bowing parameters 4 2 1 2 Hexagonal Wurtzite crystals The wurtzite crystal structure consists of two interpenetrating hexagonal close packed sub lattices Each sub lattice is shifted along the c axis by 3 8 of the cell height GaN AIN or InN exhibits a stable hexagonal wurtzite crystal structure rather than a meta stable zincblende structure In an ideal wurtzite structure the c a ratio is 1 633 The deviation from the ideal c a ratio increases as the electronegativity difference between group III atoms and group V atoms increases The c a ratio can also be correlated with the differences in electronegativity AIN has c a 1 601 and GaN exhibits a c a 1 627 while InN shows c a 1 612 The difference is attributed to the creation of the dipole resulting in polarization The unit cell of the wurtzite crystal structure is shown in the following figure O N Ga In Al Figure 2 4 Wurtzite unit
40. e n type and p type layers Because of de excitation of electrons in the conduction band when recombined with holes in the valence band photons are emitted at a wavelength function of the bandgap energy Emitted light _7 p Electrode ye p GaN contact layer Active region N contact layer Buffer layer Substrate _ 20 nm In GaN Mg 100 nm In GaN Mg WB Mee a InGaN nGaNQW2 o oo O OO GaN QWB InGaN QW 1 3 pm GaN Si 1 pm GaN ud Sapphire Figure 3 1 Left Conventional Structure 10 and Right Triple Quantum Well Structure 14 The following sub sections will discuss the principal layers of an LED structure But at first we will begin with a general overview of the characteristics of a p n junction 3 2 P N junction A p n junction is composed of one region uniformly p doped in one side and one n doped on the other side carrier concentration log scale T lt lt E field 1 Diffusion force on holes Diffusion force on electrons j 1 E field force on holes E field force on electrons 1 j Figure 3 2 Illustration of a p n junction 11 When the two regions are put into contact a diffusion of carriers occurs because the n material has a large concentration of electrons and conversely for the p material A diffusion current is then established from n side to p side for electrons and inversely for the holes The resulting diffusion current can
41. easurement erasini n e A E 49 AO Ray Diacon ARD e veiileethdec tstaiedelesthaiad veteta elec aeddleaiess 51 CHAPTER 5 TEMPERATURE DEPENDENT PHOTOLUMINESCENCE ccceeeeee 57 SoA ET a E E E E E E EE case EAE E NEO 58 D2 Oia POM pee A cestuaaachedeutecanenceaeeen 59 IE O saaceltmaati tee cuueaioanedtcnnasauitndshssasdaocusessenddtanacsauimmcneotcmeseiees 60 74 SPECULUM ACUI ON eeni a 61 39 PU ite SU ae ns ele cde ae ee ee le a 63 CHAPTER G QUICK U0 oss cechennsctadenscestnttteaalaeetennidiadend den eatin etieer eet eads T2 LLABE W een ton Meare ren Ser Renney aren Mreaeem ey cen Sen neem tie neem erat T2 G2 NONANO e a a a erated aces 74 6 3 Front panel and ex CCUM diiseni rn ee aaetuetd a aiiaaaeadscas eae 76 6 3 Connection Information Pad seseina 77 6 3 2 Sample information and Spectrometer parameters ceeeeeeeeeeeeeeeees 78 6 5 5 Mode and Range Selecion sans sesccatea thud E Sian saeea th eee eaaaeandes 79 6 3 4 Sweep Current List Current Control Tabs 0 0 ccccccsssssssseessseeeeeeeeees 79 6 3 5 DC Pulse Mod Parameters Tabs vc c2i tcccccieuseceedadssascieisecendelloadsieimeaenetsanies 80 G9 O55 PECTED S a Aaa ae aa eee ace aaa 81 OTAEV Chara reniSie e ald eta etait a 82 OS PIOCEssed Daar RESU E cens A E 82 GA Results and Compari S Olssssreni a Mada g aiviaties 82 CHAPTER CONCLUSION retann i E 86 REFERENCES cererea lc comestlehsclaate oatsnaes Saat a caeresabchaatioceditea estate td sein
42. ecause the limit of the free carrier concentration is larger with Si doping 4 x 10 cm due to relatively low ionization energy of Si donors about 20 meV compared to Ge doping P type doping in III N materials is one of the major technical challenges to overcome in order to improve LED device performance Magnesium Meg is generally used as a dopant in GaN based materials for p type layers But Mg doped GaN epitaxial layers do not have a high electrical conductivity because of hydrogen bonding for example and the doping efficiency is only around 1 due to a large ionization energy of Mg around 120 meV However a more active region friendly layer can be grown with InGaN Mg because p InGaN layer is grown at lower temperature to avoid indium in the active region to be ejected out and the Mg acceptor activation energy is lower in InGaN Mg than for GaN Mg see section 3 4 2 2 Metalorganic Chemical Vapor Deposition Metalorganic chemical vapor deposition MOCVD has been developed over the past forty years to become the dominant epitaxial materials technology for both research and production for I Nitride related materials growth This technique has enabled LEDs to become commercially viable The MOCVD epitaxial growth technology was pioneered 9 in 1968 by Manasevit et al 6 by demonstrating the heteroepitaxial growth of I V materials on insulating substrates But because of the early stage of its development the epitaxi
43. ed electrons reach the opposite extremity of the active region while a smaller quantity of injected holes have the time to penetrate into the active region resulting in an significant number of electrons in the conduction band that can t recombine Therefore these electrons propagate further and penetrate in the p type layer or even collected in the metallic contact referred to an overflow or spillover effect One important field of research is to find a way to increase the mobility of holes in the p type layer and active region in order to enhance the electron hole 30 radiative recombination One example of structure modification comes from workers at Virginia Commonwealth University who have proposed 29 to reduce the thickness of the barrier in an InGaN multiple quantum well structure from 12 nm to 3 nm allowing better hole penetration They claimed to have increased the current density of the peak external quantum efficiency from 200 A cm to 1100 A cm Another idea under investigation in our group is to optimize the electron blocking layer refer to section 3 9 3 between the active region and the p type layer that acts as a barrier to electron propagation Therefore much more electrons are retained in the active region thus enhancing radiative recombination Recently a graded electron blocking layer GEBL has been introduced 14 to overcome some issues due to a single layer EBL 3 8 3 Polarization effects Polariza
44. engths thus demonstrating the propagation of holes through the three quantum wells Therefore an improvement of hole transport may be experienced But when compared to the sample with a p GaN layer the conclusion can t be clearly extracted from the results The EL spectrum shows the emission of two distinct peak wavelengths but they correspond to the first and third quantum wells which raise questions about what has happened in the second quantum well that explains a quasi inexistent light emission from that well 33 Intensity arb unit htensity arb unit Wavelength nm Wave length nm Figure 3 14 T W MQWs EL spectrum with left p GaN and right p InGaN AMDG Internal Document More interestingly the change of In composition in the p InGaN gives rise to results that are counter intuitive The following figure shows the same structure but with different indium composition 1 5 2 and 3 in the p type layer nit Intensity arb u Intensit y arb unit nsity arb unit hte Wavelength nm Wavelength nm Wavelength nm Figure 3 15 TW MQWs EL spectrum with different In composition of the p type layer left 1 5 center 2 and right 3 AMDG Internal Document When In composition increases the last quantum well experiences an important augmentation of radiative recombination while others remain approximately of a similar relative magnitude Furthermore several batches of this same stru
45. eposition courtesy of AMDG P bonding pad deposition 4 This is the last step of the device fabrication process As stated previously the p type layer suffers from poor hole transport due to difficulties in growing highly conducting p type HI N layers Therefore improving the ohmic contact quality is of prime importance to increase holes injection into the p type layer One of the solutions is to add a layer of gold around 200 nm in top of the p spreading contact layer for bonding purpose We have to note that the gold bonding layer is not transparent to the visible wavelengths Active Region N contact P type layer P spreading Buffer Layer undoped GaN Gold Bonding Pad Sapphire Substrate Figure 3 10 Left Schematic and Right image after p bonding pad deposition courtesy of AMDG 24 3 8 Efficiency Droop LEDs require an electrical current to be injected into the structure in order to emit light The general trend is the higher the injected current is the brighter the emitted light is But for GaN based LEDs beyond a certain current density approximately 10 A cm this trend tends to degrade with increasing injection current which has been referred to efficiency droop The causes of this phenomenon are still not well understood The following figure illustrates the efficiency droop in the EL of blue LEDs grown with different p type layers but emitting at the same wavelengths The normalized integ
46. f spectrum acquisition and processed data results The benefit for the research in the AMDG was to reduce measurement time improve efficiency supply a more user friendly front panel and to enable transfer to other computers 1X INTRODUCTION 1 1 Why is the LED a new promising technology The principle of light emission exists in many different forms e g incandescence fluorescence gas discharge luminescence or solid state lighting SSL using Light Emitting Diodes LEDs Introduced as a practical electronic component in 1962 early LEDs emitted only red light but modern versions are available with emission across the visible ultraviolet and infrared wavelengths When a light emitting diode is forward biased electrons are able to recombine with holes within the device releasing energy in the form of photons This effect is called electroluminescence and the color of the light corresponding to the energy of the photon is determined by the energy gap of the semiconductor LEDs are often small in area less than 1 mm and integrated optical components may be used to shape its radiation pattern LEDs present many advantages over other light sources including lower energy consumption longer lifetime improved robustness smaller size faster switching and greater durability and reliability Light emitting diodes are used in applications as diverse as automotive lighting traffic signals full color outdoor displays and ba
47. f the material cm and V4g the Hall Voltage V In practice the Hall coefficient Ry is usually used to derive carrier concentration and mobility Vap t R eye 4 6 For n type materials the calculations are the same except that the magnitude of the electronic charge q the Hall voltage Vag and the Hall coefficient Ry are negative The measurement is carried out using an Accent HL5500PC Hall effect system where a 1x1 cm square sample is prepared by depositing an ohmic contact to each corner 50 labeled 1 2 3 4 Free standing resistivity is measured by applying a current across one pair of contacts and measuring the voltage across the other pair in the two configurations as per the following equation V3 V p 2 266 t 3 F 4 7 12 l4 where is the sample thickness is the applied current V is the measured voltage and F is a correction factor based on sample symmetry 1 for a perfect square 4 6 X Ray Diffraction XRD X rays are electromagnetic radiation with typical photon energies in the range of 100 eV to 100 keV For diffraction applications only short wavelength X rays in the range of a few angstroms to 0 1 angstrom 1 keV 120 keV are used Because the wavelength of X rays is comparable to the size of atoms they are ideally suited for probing the structural arrangement of atoms in a wide range of materials The energetic X rays can penetrate deep into the materials and provide in
48. formation about the structural properties such as composition and uniformity of epitaxial layers thickness built in strain and strain relaxation and crystalline perfection related to the dislocation density 5I PORERNE X Ray Source 26 Analyzer Crystal Four Crystal Bartels Monochromator Detector 1 Detector 2 Figure 4 8 Schematic of an XRD system 5 X rays are produced generally by either X ray tubes or synchrotron radiation In a X ray tube which is the primary X ray source used in laboratory X ray instruments X rays are generated when a focused electron beam accelerated across a high voltage field bombards a stationary or rotating solid target As electrons collide with atoms in the target and slow down a continuous spectrum of X rays are emitted which are termed Bremsstrahlung radiation The high energy electrons also eject inner shell electrons in atoms through the ionization process When a free electron fills the shell a X ray photon with energy characteristic of the target material is emitted Common targets used in X ray tubes include Cu and Mo which emit 8 keV and 14 keV X rays with corresponding wavelengths of 1 54 A and 0 8 A respectively In recent years synchrotron facilities have become widely used as preferred sources for X ray diffraction measurements Synchrotron radiation is emitted by electrons or positrons travelling at near light speed in a circular storage ring These powerful sour
49. gas Then the helium gas is sent to the cold head to cool down the cold finger where the samples are attached to The system is based on Gifford McMahon refrigeration cycle where the helium gas absorbs heat by expansion on heat stations in the cold head The cold created by this mean is transferred to the sample holder in the cold finger A vacuum turbopump is used to create vacuum in a vacuum jacket where the sample holder is located in order to prevent thermal transfer and to avoid the creation of ice from air moisture inside the system A heater plugged in the sample holder and controlled by a temperature controller along with a GaAs diode sensor enables a control and monitoring of the sample temperature Thanks to this system the spectrum acquisition is done from 10 K to 310 K with a 20 K Step 5 4 Spectrum Acquisition Because the emission from the sample is isotropic a focus system is required to focus the light into a small spot that enters the monochromator The monochromator Triax 190 functioning is based on diffraction using a triple axis grating turret fitted with a 61 visible blaze diffraction grating having 1200 grooves mm to spatially separate the colors of light The optical system is designed according to the Czerny Turner design refer to Figure 5 3 The broad band illumination source A enters the monochromator at the entrance slit B The amount of light energy available for use depends on the intensity of
50. h the p type layer by creating a high potential barrier between active region and p type layer The thickness of the EBL is typically between 5 to 20 nm 19 The EBL is usually made of AlGaN because it has a larger energy bandgap than the GaN barriers and InGaN wells in the active region However better performance has been experienced using InAIN rather than AlGaN The reasons and results are discussed in section 3 9 Another way of improving the electron blocking mechanism may be to use a graded electron blocking layer GEBL A research group in Taiwan from National Chiao Tung University has developed a GEBL that reduces efficiency droop compared to devices with a conventional AlGaN EBL 14 The efficiency droop for the GEBL LED at an injection current density of 200 A cm is only 4 from the peak value compared to 34 with a conventional single EBL The following figure shows that the originality of this approach is a dissymmetry about its effect on the energy bands The conduction band quantum barrier is increased which reduces electron propagation while the barrier in the valence band is decreased which may improve hole propagation Conventional GEBL Hole concentration cm log Electron concentration cm log Figure 3 4 a hole and b electron concentrations distribution of conventional and GEBL LEDs at a current density of 100 A cm 14 20 Output Power mW Conventional GEBL 0
51. his issue The IQE is calculated by the following equation Integrated Intensity T1 Integrated Intensity T2 IQE g T1 7T2 5 3 Where 77 7 and T2 are temperatures and the intensity is integrated from 350 to 650 nm In our case the temperatures chosen for the IQE calculation are the minimum 10 75 K and the maximum 310 K The following figures show the integrated intensity function of temperature and the corresponding IQE 69 Integrated Intensity vs Temperature 1 500E 06 5 1 400E 06 1 300E 06 4 y 2 080 80x 1 441 407 38 R 0 95 1 200E 06 4 1 100E 06 4 1 000E 06 9 000E 05 8 000E 05 7 000E 05 4 Integrated intensity a u 6 000E 05 Temperature K Figure 5 8 Integrated intensity function of temperature for IQE calculation Raw IQE 54 68 IQE Linear 56 12 Figure 5 9 IQE results for the sample no 2 2506 3 As expected the raw IQE using raw integrated intensity data and the linearized one are different but the difference is quite insignificant However one of the problems to overcome before reaching this result was to improve the thermal contact between the sample and sample holder because the difference was previously higher around 8 The previous samples were mounted using copper tape instead of grease The following figures show the results with copper tape 70 Integrated Intensity vs Temperature 1 800E 06 7
52. ing a digital response The spectrometer is controlled by a LabVIEW program QuickTest 2 0 More detail about this program will be provided in Chapter 6 4 2 Photoluminescence Photoluminescence PL is based on the same principle that EL but the difference is no carriers are electrically injected into the device An optical source usually a laser acts 41 as an optical pump that excites electrons from the valence band to the conduction band through the absorption of the energy provided by the incoming beam of photons Photon Absorption Conduction Band Rapid Decay Incoming photon from optical source wi Photon emission we E 1 Valence Band Figure 4 3 Illustration of photon generation by Photoluminescence PL measurements permit the assessment of the performance of the active region In contrary to EL measurements metallic contacts and a p type layer are not mandatory because no carriers are electrically injected which gives us a way to specifically assess the quality of the active region as well as the MQW bandgap energy The optical pump wavelength has to be shorter than the target wavelength of the emission in order to excite the electrons into the conduction band However a too short wavelength will enable radiative recombination in the n GaN layer which is not of interest for this measurement Furthermore the optical pump wavelength should be carefully chosen to enable radia
53. ing sets the boxcar smoothing width a technique that averages across spectral data This technique averages a group of adjacent detector elements A value of 5 for example averages each data point with 5 points to its left and 5 points to its right The greater this value the smoother the data and the higher the signal to noise ratio 78 If the value entered is too high a loss in spectral resolution will result The S N will improve by the square root of the number of pixels averaged 6 3 3 Mode and Range Selection Mode Selection selects between DC Pulse and Automatic modes When DC is selected only the DC Mode tab s parameters can be modified and inversely for the Pulse mode However when Automatic is selected both tabs are available for modification Automatic mode enables the software to switch between DC and Pulse automatically The Automatic Threshold mA control indicates the current threshold from which the switch occurs from DC to Pulse mode The Wavelength Range selects the spectrum saved and displayed UV 200 400 nm Blue 350 500 nm Green 400 600 nm Default 300 nm to 700 nm or Full 200 m to 1100 nm spectrum The Execution button enables to start the measurement 6 3 4 Sweep Current List Current Control Tabs Sweep List Start Current mA Current List 10 Stop Current mA pun Increment mA mam Actual Current mA Actual Current mA _ fso so Figure 6 5 left S
54. materials because it 1s grown under conditions which are more active region friendly than p type GaN InGaN is grown at a lower temperature than GaN which prevents damage to the InGaN quantum wells in the active region Indeed a high temperature growth can be responsible for indium to be ejected out from the crystalline structure of the active region resulting in crystal quality degradation Besides the Mg acceptor activation energy is lower for InGaN Mg than for GaN Mg which permits to obtain higher free hole concentration at 300K above 10 cm Furthermore using InGaN p type layer may reduce the contact resistance due to a smaller bandgap energy 12 17 However the drawback is a higher lattice mismatch because the last layer in the active region 1s GaN which results in increasing polarization effect A typical thickness of our p type layer in the LED structure is around 120 nm 3 5 Active Region The active region is the core of the LED structure and this region in the LED determines the way that the photons created by electron hole radiative recombination are emitted Two types of active region are typically employed single quantum well SQW and multiple quantum well MQW In our structure the MQW is composed of two different III N materials usually GaN as the quantum well barrier and InGaN as the quantum well in which radiative recombination occurs The thickness of the whole active region is around 60 nm one
55. measured Voltages V characteristic is done automatically and saved into the data file instead of writing it down at each measurement Therefore the benefit for the group is a reduction of the measurement time improvement of repeatability and the results are more accurate due to decrease of heat effect as the measurement process is automatically controlled Finally the program is robust and is easily transferable to another computer 6 3 QuickTest 2 0 front panel and execution The front panel is the window the user interacts with It contains the input parameters and displays the EL spectrum I V characteristic and processed data result of the LED sample under measurement The following figure is a screenshot of the front panel 76 KEITHLEY INSTRUMENTS INC MODEL 2430 Pulse Mode t a pr ON time 5 486 000 483 400 481 900 480 900 484 731 482 453 480 174 _ 479 718 29 798 0 1 3 s offset 28 342 28 299 28 682 29 176 9 630 9 2 89 630 91 572 92 458 92 957 1 r 7483 834 26685 69 35629 55 44104 37 86 033 125 386 162 154 196 259 4 48 5 06 5 61 6 13 2 9 OFF time s 299 587 619 704 942 565 1254 830 1550 221 1 8 lt Figure 6 4 Front Panel of QuickTest 2 0 6 3 1 Connection Information Pad The GPIB address is the identification number of the source meter in the GPIB bus It is required for the program to know which instrument it has to set a communication with
56. mediated by a scattering mechanism which provides additional momentum and enables Auger transitions to connect to a broader range of final states This process is important in the nitrides and can account for at least some of the efficiency droop effects in nitride LEDs 33 29 One scattering mechanism that assists IAR is the electron phonon interaction which is particularly strong in the nitrides 34 Another scattering channel is introduced in the active region made of InGaN layers by the alloy induced symmetry reduction Charged defects may also scatter carriers and cause Auger recombination Phonon and alloy assisted Auger processes are strong and cumulatively account for a sizeable Auger coefficient 33 Thus the indirect Auger recombination is currently one of the first candidates as a contributor to efficiency droop due to theoretical and experimental results 33 that show a high value for the indirect Auger recombination coefficient 3 8 2 Current Overflow Current overflow is a carrier loss mechanism in which electrons recombine outside the active region either radiatively in which case the recombination occurs at an undesired wavelength or nonradiatively or collected by the metallic contact The reason is that electrons have an effective mass seven times lighter than holes in GaN related materials which allows them to propagate faster in the crystal Therefore when a current is applied to the LED structure the inject
57. mileds are among those that believe Auger recombination is the source of the efficiency droop effect 28 while others from Virginia Commonwealth University criticize this idea and argue the difference in effective mass between electrons and holes is mainly responsible for the efficiency fall off 29 In a different route 20 Rensselaer Polytechnic Institute s researchers are convinced that the polarization effect in the Quantum Well QW enhances the leakage of injected electrons into the p type layer which according to them is the primary mechanism that leads to an efficiency droop The aim of this thesis is to describe different ideas that the AMDG group is investigating to reduce the efficiency droop by improving LED structures The studies carried out in our group are focused on the reduction of the current overflow effect by using an optimized electron blocking layer increasing hole injection and transport efficiencies by 26 optimizing the p InGaN layer and the InGaN MQW active region and polarization effect by engineering the crystal growth of the multiple quantum wells on the active region The following sub chapters will introduce the most important contributors to the efficiency droop that are currently under intense investigation Auger recombination current overflow and polarization effects 3 8 1 Auger Recombination Auger recombination is a nonradiative mechanism in which the electron hole recombination is
58. ms fe Ref amp ole PD _ am a fero Delay Pulse width delay used to achieve pulse width setting 80us Minimum pulse width overhead Meas Sig Signal measurement 9ms Minimum gure of time overhead Meas Ref amp Zero Reference and zero measurement PD Pulse delay setting used to determine time between pulses Figure 6 7 Pulse configuration in the source meter Keithley 2430 User s Manual The Trigger Count is the number of pulses outputted for each current level The Pulse Width is the width of the pulse On Time that has to be between 0 15 ms and 5 ms The Pulse Delay is the delay during Off time and has a maximum value of 9999 999 s The Duty Cycle is calculated by the program according to the Pulse Delay and Pulse Width settings 6 3 6 Spectrum Graphs This graph displays the spectrum in two steps A first graph displays the current spectrum under acquisition for each current At the end an overlaid spectrum graph shows all plotted spectrum in one graph as illustrated in the front panel image The x axis is automatically sized according to the wavelength range selected However the min and max values can be changed at any time and the plot resizes automatically 81 6 3 7 I V Characteristic It is a graphical display of the I V result that is updated on real time for quick assessment of the electrical sample performance 6 3 8 Processed Data Result This table summarizes different spectrum charac
59. n fact the previous version QuickTest was programmed with an older version of LabVIEW and was designed for various applications in one program Besides the program suffered from containing an excessive number of programming structures which slowed down the measurement The following figure shows the EL measurement station setup 74 LED Sample Figure 6 3 Illustration of the EL measurement station The basic functioning of the EL measurement station is that a current is applied to the LED sample through metallic contacts on the sample surface Therefore photons are emitted due to radiative recombination and are transported from the backside of the sample through a fiber optic to the spectrometer Then the digital data from the spectrometer is sent to the computer via USB The program receives and displays the raw data and saves it to a file along with other useful processed data QuickTest 2 0 has a lot of advantages compared to the first version At first the execution of the program is faster it takes seconds instead of minutes to measure a sample It can control the Keithley source meter and synchronize its execution with the spectrometer 75 while these two manipulations had to be done manually previously The program displays the EL spectrum in real time and a preview of the overlaid spectrum There is no limit of measurements per session compared to nine for the previous version Finally the acquisition of the
60. n has also been investigated The main problem during device fabrication is to obtain a low ohmic contact resistance between the metal and the p type layer The previous fabrication process resulted in a contact resistance of 3 23e 3 Q cm but the contact J V curve was that of a Schottky barrier instead of an ohmic contact The first study concerns the thickness of the metal layer composed of nickel and silver By decreasing the nickel layer thickness from previously 5 nm to 2 nm while keeping the silver one at 100 nm the contact resistance has been lowered from 3 23e 3 Q cm to 1 23e 3 Q cm Furthermore the contact is now ohmic The second improvement currently under investigation is related to the annealing temperature where an optimal temperature of 550 C instead of previously 500 C may permit a further decrease of the metallic contact resistance 38 4 EQUIPMENT FOR MEASUREMENT PURPOSE All the results and improvements have been made possible thanks to the analysis of measurement data AMDG has not only the equipment to grow materials but also the capability of carrying out almost all the required measurements on our LED devices from material characterization to spectrum acquisition The following list is not exhaustive but gives an overview of the most important ones 4 1 Electroluminescence Electroluminescence EL is the phenomenon of light emission by radiative recombination due to injected carriers from an electrical cu
61. n pattern are directly related to the atomic distances For a given set of lattice planes with an inter plane distance of dak the condition for a diffraction peak to occur can be simply written using the Bragg s law 2 Anke sin Op n 4 8 53 For a Bravais Lattice 4 9 d nx 3 l c2 1 Afh hk k where is the wavelength of the x ray Og the scattering angle n is an integer representing the order of the diffraction peak hkl are the Miller indices of the crystal and a and c Bravais lattice constants 2dsind A Hinan Braggs Law Figure 4 9 Illustration of the Bragg s law in the case of a 2D lattice plan 39 Thin film diffraction is used to characterize thin film samples grown on substrates There are several special considerations for using XRD to characterize thin film samples At first reflection geometry is used for these measurements as the substrates are generally too thick for transmission Secondly high angular resolution is required because the peaks from semiconductor materials are sharp due to very low defect densities in the material Consequently multiple bounce crystal monochromators are used to provide a highly collimated X ray beam for these measurements For example in the Philips X PERT MRD used in our lab 54 a four crystal monochromator made from Ge 220 is used to produce an incident beam with 0 00001 of angular resolution
62. ning the red shift phenomenon In case of InGaN 43 eV _ 4 Eg reo 2 55eV a 8 32x 10 y and p 835 6 K Hence amn E hxc o E m aT go T E Ir K T Peak Wavelength Shifting vs Temperature g S ON 3 oO z Temperature K Figure 5 5 Red a Simulation using Varshni formula and Blue 0 measured peak wavelength shift 65 The difference between the Varshni simulation and the measured shift 1s due to InGaN localization combined with inexact Varshni coefficients as they depend of In composition in the alloy and have to be determined experimentally InGaN localization plays a role in the shift of the central wavelength as well as broadening the spectrum Indium has a tendency to not be uniformly distributed in InGaN material but rather forms clusters and temperature gradients across the substrate can lead to segregation of indium which causes localized potential minima 35 Since higher indium content means lower bandgap a non uniform distribution of energy potentials will form across quantum well layers and photons of different energy and thus different wavelength will be emitted from different parts of the wafer At a temperature lower than 70 K the carriers have not enough energy to pass over potential fluctuations therefore the radiative recombination occurs at the same position where electron hole pairs have been generated But with increased temperature until around 150
63. not build up indefinitely because an opposing electric field is created at the junction as electrons leave behind uncompensated donor ions Nq in the n side and holes leaving uncompensated acceptors Nx This electric field is in the direction opposite to that diffusion current for each type of carrier Therefore the field creates a drift component of the current opposing the diffusion 15 current The result is a net current equals to zero at the equilibrium as well as a built in voltage Vo When the p n junction is forward biased the potential barrier that exists at the space charge region is lowered due to an opposite applied electric field compared to the built in field The consequence is the flow of carriers is facilitated because of a low potential barrier A current induced by the flow of the majority carriers is then created 3 3 N Type Layer The n type layer of a GaN based LED structure is usually composed of GaN doped with silicon The choice of using silicon donors comes from the small ionization energy of donor bound electrons A small ionization energy means that silicon donors can easily give off electrons to the conduction band resulting in a higher concentration of free electrons Free electron concentrations can be controllably changed from 10 to 2x10 cm by varying the flow rate of silane Si1H4 the precursor used for Si doping of GaN in MOCVD Si substitutes for a Ga atom in the lattice and provides a loo
64. ns E T1 Figure 6 5 left Sweep Current Tab and right List Tab 0 ec cccsseeeeeeeeeeeeeeees 79 Figure 6 6 left DC Mode Tab and right Pulse Mode Tab ccccccsseessseeeeeeeeeees 80 Figure 6 7 Pulse configuration in the source meter 0 0 eeecccccceeeeeeeeeeseeeseeeeeeeeceeeeees 81 Figure 6 8 Variance of intensity at different currents cc cccectrnteeeeeeteeseeeeeeeeeeeeees 83 Figure 6 9 Variance of the processed data result at different current levels 84 Vil SUMMARY The objective of this thesis is to discuss the solutions investigated by AMDG Advanced Materials and Devices Group to reduce the efficiency droop effect that occurs in III Nitrides Light Emitting Diodes LEDs when driven at high injection current densities The efficiency droop refers to a decrease of the LED light emission efficiency when increases the current density from low values 10A cm to higher values gt 100A cm Many scientific papers have been written about the possible reasons for this phenomenon Therefore this thesis will discuss the different effects suspected to contribute to the droop and discuss LED structure modifications studied by Dr Dupuis research group to reduce their impact In addition to a description of a conventional LED structure a discussion of the device fabrication process will be provided including the solutions investigated in our group to improve LED performance Because measureme
65. nt is critical to our studies a description of the equipment used by the AMDG will be provided e g the Electroluminescence EL and Photoluminescence PL test stations Atomic Force Microscopy AFM for surface topology TLM for metallic contact resistivity X Ray diffraction for crystal quality and epitaxial layer structure and Hall Effect measurement for doping concentration characterization and material resistivity Because the IQE gives us a direct assessment of the active region s crystal quality the setup and operation of a new Temperature Dependent PL TD PL system to measure the Internal Quantum Efficiency IQE was the main focus of this research The External Quantum Efficiency EQE is measured using electroluminescence measurements The Vill EL measurements involve the acquisition of the emitted light spectrum along with different processed data such as the Full Width at Half Maximum FWHM of the spectral intensity the peak wavelength output power etc which allows a comparison of the different LED structure performances Within this work a new LabVIEW program called QuickTest 2 0 has been developed in order to automate the instrumentation setup and improve both the speed and accuracy of EL acquisition A brief description of the G language used by the LabVIEW software will be provided along with the objective and motivation for upgrading the program the general features of the program and a comparison o
66. nt under high vacuum The electron beam causes a region in the target metal source to melt and atoms from the target to transform into a gaseous phase Then these atoms travel through the vacuum and precipitate into solid form coating the sample surface with a thin layer of metal The metal used for n contact is a superposition of different layers of metal 22 Titanium 221 A Aluminum 905 A Titanium 269 A Gold 460 A Then a rapid thermal annealing RTA is carried out in order to improve the contact quality at a temperature of 700 C Active Region P type layer Buffer Layer undoped GaN N contact metal Sapphire Substrate Figure 3 8 Left Schematic and Right image after n contact metal deposition Courtesy of AMDG P spreading metal deposition 3 The third step is the deposition of a current spreading layer on top of the p type layer The use of a current spreading layer is crucial because it permits a better distribution of holes over the p type layer The process uses photolithography and the deposition of Nickel 60 A Gold 60 A layers if transparent metal is desired or Nickel Silver for a reflective contact using e beam evaporator Then an RTA process is carried out at 500 C 23 N contact Active Region N type layer P type layer P spreading Buffer Layer undoped GaN Sapphire Substrate Figure 3 9 Left Schematic and Right image after p spreading metal d
67. ntum barrier which is typically used in this case Even if the GaN Mg layer is grown at high temperature the damage on the indium in InGaN quantum wells for blue LEDs is 32 not strong enough to degrade the overall performance 35 But when it comes to green wavelengths the growth temperature of the p type layer greatly impacts the quality of the active region due to a high indium composition of the quantum wells leading to low quantum efficiency Therefore the idea is to use InGaN p type layer in order to lower the growth temperature referred as active region friendly layer Thus the objective of this research is to determine the difference of light emission efficiency between using p GaN or p InGaN with different In compositions To keep track of hole propagation in the active region a Triple Wavelength MQW has been fabricated and is composed of three quantum wells of different indium compositions in order to emit at three different wavelengths 10 for 425 nm 17 for 460 nm and 23 for 520 nm EL measurement is used to monitor hole propagation in the active region by showing emitted wavelengths So far there is no clear evidence of improvement but points out interesting problems that are still not explained In the following figure a Triple Wavelength MQW structure was fabricated with a p GaN layer left and p InGaN layer right The EL spectrum of the structure with a p InGaN layer shows an emission of three peak wavel
68. om 460 nm to 530 nm m flmA Variance hm Wavelength nm Wavelength nm 51837 58 54102 4003 446 ME W 9566 Ww MI W7 MM 48428 Figure 6 8 Variance of intensity at different currents In case of low current injection lt 10 mA the difference is inferior or equal to 12 for the spectrum of interest and less than 6 at high current injection gt 10 mA This gap between low and high current injection is due to the fact that at low injection the intensity is very low and thus a small difference will lead to a higher variance 83 However the most important data used for comparison of samples is the processed data and more specially the peak wavelength dominant wavelength and FWHM The following figure shows the variance for the different processed data Dominant Wavelength 0 05 0 045 0 035 0 025 0 20 40 60 Peak Wavelength 100 0 05 A __20 40 60 __ 80 100 Saturation 100 FWHM 0 20 40 60 80 100 X axis Current mA Y axis Variance 20 Luminance 40 60 80 100 Power Figure 6 9 Variance of the processed data result at different current levels As shown on these graphs the variance for the peak wavelength is less than to 0 3 1 5 for the FWHM and 0 05 for the dominant wavelength which is more than acceptable The l
69. on efficiency In addition close packed wafer configuration results in high area utilization These two factors produce high precursor utilization efficiency 11 The three zone heater system creates the temperature uniformity via modification of the temperature profile resulting in higher yield The figure 2 6 illustrates the basic diagram of an MOCVD gas inlets gas inlets growth chamber Hydrogen Carbon e Nitrogen Gallium Hydrogen Carbon a b gas inlets gas inlets growth chamber Hydrogen Carbon e Nitrogen Gallium e Hydrogen c Figure 2 5 Illustration of MOCVD epitaxial growth process a gas input b pyrolysis c diffusion and d surface reaction 9 12 Mi kd z NI I E Ke aal 4s N Si Source Gas Glovebox SiH Carrer Line Pyrometer Load Interfereometer Chamber EER Carrier ihi A Gas j i I A Mass Flow D amp Controller Filter Pressure Controller Control Pump Valves Exhaust Three Way Valves Metal Organic Source Figure 2 6 Basic diagram of an MOCVD reactor 5 13 3 LED DEVICE BASICS 3 1 Principle Light Emitting Diodes are a special type of a p n junction A current is injected in the device where electrons flow from the n type layer to the p type and inversely for holes Then electrons and holes recombine in the active region of the LED located between th
70. rated intensity plotted on the Y axis is the result of the integration of the EL intensity over a wavelength range usually from 350 nm to 700 nm for blue and green LEDs and then normalized according to the highest integrated intensity s value Maximum Integrated Intensity O k La Lee 4 fa m TW LED wi p GaN TW LED wi pdn 00s Ga Sees i TW LED wi p n GaN N 0 0 Normalized Integrated Intensity arb unlit 50 a zai 100 150 200 250 300 350 400 Current Density Alc m Figure 3 11 Illustration of the Efficiency Droop AMDG Internal Document 25 Many different contributions to droop have been identified explained and discussed in the literature current overflow 15 poor carrier injection efficiency 16 17 18 polarization fields 19 20 Auger recombination 21 junction heating 22 carrier delocalization from quantum dots 23 exciton dissociation 24 high plasma carrier temperature 25 and quantum confined Stark effect 26 Among these nonradiative recombination Auger effects current overflow or current spillover and polarization effects represent perhaps the main mechanisms responsible for the observed efficiency degradation However at this time no one can undoubtedly state which one dominates Various experiments have been developed in order to observe the impact of one effect over the others To illustrate this controversy 27 researchers from Philips Lu
71. rrent The electrons injected to the n type layer and the holes to the p type layers propagate through their respective layer by diffusion When they reach the active region the electrons in the conduction band recombine with holes from the valence band The energy lost during the de excitation of the electron creates a photon The emitted photons propagate with a wavelength associated to the bandgap energy see equation 2 1 gt Emission of a Photon p type layer Figure 4 1 Illustration of radiative recombination process 39 Analyzing EL spectrum enables us to compare LED performance Even if the light intensity is displayed in arbitrary units comparing useful data such as peak intensity peak wavelength full width half maximum among others is of prime importance MQWs increase the probability of recombination therefore the major peak will be ideally at a wavelength corresponding to the energy bandgap of these wells However the analysis of the EL measurement can display emission at other wavelengths which indicates an overflowing of the wells or recombination at some other site in the material Defects create nonradiative recombination centers and will thus degrade the EL performance of the device The EL equipment we are using is composed of a stage that supports the LED sample a fiber optic cable and two pins connected to a current source Keithley 2430 The current source generates an electrical current in D
72. rter planet revolution by providing an alternative to inefficient incandescence or phosphorescence lighting not only in term of performance but also of cost But before thinking about deploying this technology we still have to bridge the green gap due to the efficiency reduction effects by understanding the mechanisms that prevent high brightness green LEDs from being fabricated A description of the main effects suspected to be responsible for the efficiency droop common to all HI N LEDs which are Auger recombination effect current overflow and polarization effect have been provided III Nitrides are used for blue and green LEDs as they exhibit a direct bandgap and can emit over the entire visible spectrum A brief discussion about the proprieties of III Nitrides was given in order to understand the performance of M N LEDs along with a description of the MOCVD epitaxial growth technique that is mostly used for GaN related material growth The discussion of Mg doping of GaN crystals gave a first understanding of the problems related to the p type layer growth and introduced the advantage of growing a p InGaN layer due to a lower growth temperature required but exhibits a higher lattice mismatch with the active region than p GaN layer The issues with LED structure growth and the solutions the group is investigated are discussed in the LED device basics chapter where the characteristics of the n type p type and active region are presented
73. seaett 88 LIST OF FIGURES Figure 1 1 External Quantum Efficiency for the visible spectrum cccseeeeeeeeeeees 2 Figure 2 1 left direct vs right indirect bandgap cccccccccccssssesseeeeeeeeeeaaesseeeeeeeeeeaas 5 Figure 2 2 Spectrum of emission in function of In Al and Ga concentrations 6 Figure 2 3 InAIN and InGaN Bandgaps with different bowing parametets 5 7 FLU 2a W rze unt Ces 8 Figure 2 5 Illustration of MOCVD epitaxial growth process ccccssssseseeeeseeeeeeeeeeeeees 12 Figure 2 6 Basic diagram of an MOCVD reactor sienna diwsisees coeadedicecindintieendecauiestecereaviecnds 13 Figure 3 1 Conventional Structure and Triple Quantum Well Structure 14 Ficure 3 2 Mustratom ol ap n JUN CHOON e ccieeen eee if Fis re 3 Tlustratiomot a MOW StUCtUre to cii t225c5 acta wcatabetensierdis a 19 Figure 3 4 Hole and electron concentration of conventional and GEBL LEDs 20 Figure 3 5 Output power function of current density for conventional and GEBL LED 21 Pigure 5 0 Tlustration or the mesa ctC nine iste wtick cesta esseadsiccsacdctetaheolouse estan icasacares 22 Figure 3 7 Image of the sample surface after mesa etching cc cceeeeeeseseeeeeeeeeeeeeeeeeees 22 Figure 3 8 Left Schematic and Right image after n contact metal deposition 25 Figure 3 9 Left Schematic and Right image after p spreading metal deposition 24
74. sely bound electron The ionization energy is on average 22 meV In the AMDG MOCVD systems the optimal flow rate of silane during GaN Si growth is 12 sccm and the resistivity of the resulting GaN Si epitaxial layer is 0 004 Ohm cm The growth of the n type GaN layer is nowadays well mastered and produces a good crystal quality 16 3 4 P Type Layer As previously indicated the p type layer is currently one of the most critical issues in the improvement of LED performance Magnesium Mg is the most commonly used p type dopant in GaN based materials grown by MOCVD using bis cyclopentadienyl magnesium Cp2Mg as precursor But Mg has a tendency to bond with atomic hydrogen derived from the pyrolysis of NH3 during growth in the crystal which makes the Mg acceptors inactive as a dopant in this case Some methods have been developed in order to break these hydrogen bonds such as Low Energy Electron Beam Irradiation LEEBI or thermal annealing But even then the doping efficiency is quite low 1 because of the large ionization energy of Mg acceptor states around 120 meV The resistivity of GaN Mg in our LED structure is 1 6 Ohm cm which is around 300 times higher than n type layer 0 004 Ohm cm Therefore a loss of injected holes occurs as they recombine nonradiatively and thus dissipate the energy as heat rather than contributing to the radiative recombination in the active region P type InGaN is one of the most promising
75. t near its resonance frequency by a small piezoelectric element mounted in the AFM tip holder similar to non contact mode However the amplitude of this oscillation is greater than 10 nm typically 100 to 200 nm Due to the interaction of forces acting on the cantilever when the tip comes close to the surface Van der Waals force dipole dipole interaction electrostatic forces etc cause the amplitude of this oscillation to decrease as the tip gets closer to the sample An electronic servo uses the piezoelectric actuator to control the height of the cantilever above the sample The servo adjusts the height to maintain a set cantilever oscillation amplitude as the cantilever is scanned over the sample A tapping AFM image is therefore produced by imaging the force of the intermittent contacts of the tip with the sample surface 46 5 0 nm 2 5 NM 0 0 nm Digital Instruments NanoScope Scan size UM Scan rate 0 7535 Hz Number of samples 512 Image Data Height Data scale 5 000 nm 0 0 0 25 0 50 0 75 1 00 pm 2 1299 5 1x1 z 5nm Rq 0 145nm 212995 mm2 Figure 4 5 AFM Image of a GaN surface with scale scan parameters and data result AMDG Internal Document The AFM used by the group is a Veeco Dimension 3100 scanning probe microscope operated in tapping mode High resolution scans are obtained at 5 x 5 um This equipment is available in the MiRC 4 4 Transfer Length Measurement Transfer Length Measurement TLM is
76. terials Sci and Eng vol 13 pp 1 56 1994 42 Monochromator Wikipedia 2010 43 Y H Cho G H Gainer A J Fischer J J Song S Shaped Temperature Dependent Emission Shift and Carrier Dynamics in InGaN GaN Multiple Quantum Wells Appl Phys Lett vol 73 pp 1370 1998 44 S D Lester F A Ponce M G Craford D A Steigerwald High Dislocation Densities in High Efficiency GaN Based Light Emitting Diodes Appl Phys Lett vol 66 pp 1249 1995 45 F Haugen Introduction to Simulation with Control Design and Simulation Module in LabVIEW 8 6 Tech Teach March 2009 46 ME 295 Connect LabVIEW San Diego State University 2011 47 J Y Tsao M E Coltrin M H Crawford J A Simmons Solid State Lighting An Integrated Human Factors Technology and Economic Perspective Proc of IEEE 2009 48 C Wetzel T Detchprohm Closing the Green Gap in LED Materials Future Chips Constellation Rensselaer Polytechnic Institute 2009 91
77. teristic W Wavelength Int Integrated peak wavelength FWHM power voltage etc The table size is adjusted automatically according to the number of current values entered 6 4 Results and comparison The primarily requirement that the QuickTest 2 software has to meet is that the measurement results from the new program have to be comparable to the previous version Indeed the prominent objective of EL measurement is to compare LED samples performance based on arbitrary intensity units for instance the intensity scale Therefore if the light acquisition process differs there is no possibility to compare recent samples to the previously measured as the arbitrary units changed The spectrometer requires a lot of different inputs to work so if one of them is different from the previous version the result will be different as well Thus the idea was to keep the core of the previous program which is how the spectrum is acquired by the spectrometer but change the way it 1s processed and displayed by the 82 Variance program along with the programming structures that is responsible for the program execution In order to compare the results a green LED sample has been measured using the two versions of the program at different current levels from 2 mA to 10 mA 1 mA step and from 10 mA to 80 mA 10 mA step The following figures show the variance of intensity at different current levels The spectrum of interest is fr
78. th whom I worked closer and have shared unforgettable memories as good friends It is these colleagues who are responsible for the success of this group and who have shown me what hard work 1s Finally my gratitude is also extended to our collaborating professors Dr Shyh Chiang Shen and Dr Douglas Yoder to kindly serve as thesis committee members They are taking on their invaluable time to review this thesis 111 TABLE OF CONTENTS ACENOWLEDCGEMEN os sas ees ashes ac faa nana als dae acien sec enone tee aka cae i B Bo Ba 6 ed GG WI Ged ls eer ene ce eer te ee nen te nate eS rE Tene Pe re teNreP Rn eee Ne PES So vi SOMN ESE A A N vili CHAPTER INTRODUCTION cocisieenniin a 1 1 1 Why is the LED a new promising technology seeseeesssssssssssssssseseeerersssssssssssssssseeeeeeee l LOVE Wy OG e N N 3 CHAPTER 2 CRYSTAL GROWTH BASICS uirceinean a esaneenedescloniiaaneniuents 4 EAE Nde Mate Tal S E 4 2 1 1 Generalities direct and indirect bandgaps ccccccccecsssssssseessesseeeeeeees 4 2 1 2 Hexagonal Wu rtzite crystal aueren e e e E E Bucscees 7 2k IOpIMG GEGIN eie e a A aver at eticislugans 9 2 2 Metalorganic Chemical Vapor Deposition eeeseessssssssssseeeeeeresssssssssssssssseerereessssssss 9 CHAPTER 3 LED DEVICE BASICS croirai N 14 PEER ID Ea Ans nt sechandanieae tial enan ee se RU sectaedaieaetat sane 14 PEDT 9 0 0 Meee we er art oeei ee oa rtm Bena tn nie NY Nan Bre To nr ee nS ge 15 Bed NET Vipe We
79. the source in the space defined by the slit width x height and the acceptance angle of the optical system The slit is placed at the effective focus of a curved mirror the collimator C so that the light from the slit reflected from the mirror is collimated focused at infinity The collimated light is diffracted from the grating D and then is collected by another mirror E which refocuses the light now dispersed on the exit slit F At the exit slit the colors of the light are spread out Because each color arrives at a separate point in the exit slit plane there are a series of images of the entrance slit focused on the plane Because the entrance slit is finite in width parts of nearby images overlap The light leaving the exit slit G contains the entire image of the entrance slit of the selected color plus parts of the entrance slit images of nearby colors A rotation of the grating causes the band of colors to move relative to the exit slit so that the desired entrance slit image is centered on the exit slit The range of colors leaving the exit slit is a function of the width of the slits The entrance and exit slit widths are adjusted together The resolution of the instrument is determined by measuring the FWHM which is inferior to 0 2 nm in our system 62 Figure 5 3 Illustration of a Czerny Turner design 42 Then a CCD detector transforms the incoming light into an analog voltage As very low thermal noise is require
80. tic Or any XRD System E E 52 Figure 4 9 Illustration of the Bragg s law in the case of a 2D lattice plan 54 Figure 4 10 Example of data result from XRD superlattice measurement 56 Figure 5 1 Schematic diagram of the TD PL Setup seeeeeesssesssssssssseeeerrrreessssssssssssseeeee 59 Figure 52 Picture of a HeCd Series 14 Tase cereais E ts odeneasudetensaes 60 yi Figure 5 3 Illustration of a Czerny Turner design ccccccccccccssssesseececeeceeeeeeesseceeeeeeeaas 63 Figure 5 4 Top TD PL and bottom intensity function of temperature of the sample 64 Figure 5 5 Simulation using Varshni formula and measured peak wavelength shift 65 Figure 5 6 Illustration of indium localization effect i cccccccnnneteeeseeseseeeeeeeeeeeeeees 67 Figure 5 7 FWHM result of the sample no 2 2506 3 oo cccccnnnnteeeseseeeeeeeeeeeeeeeeees 68 Figure 5 8 Integrated intensity function of temperature for IQE calculation 70 Figure 5 9 IQE results for the sample no 2 2506 3 oo cceecccccccccceeeeeeeetesseseseeeeeeeeeeeeeees 70 Figure 5 10 Example of Top a TD PL and bottom IQE result using copper tape 71 FPieure 6 1 Example Ol a Oni pane lorire E weaeenaaeats 73 Figure 0 2 Exdmpleol a DIOCK dasri ies E 74 Figure 6 3 Illustration of the EL measurement station sssseesssseeeeeeeeerereesssssssssssseseeees 75 Fieure 6 4 Front Panel Of Quick Lest 2 0 ccoicictsenvtcnas tudinal se
81. ting in allowing energy states in the forbidden band where electrons can be trapped and emit photons at different wavelengths This can be observed in the emission spectrum of an LED via a broadening of the linewidth from the central peak wavelength and calculated by determining the distance in nm between the right and left edges of the spectrum at half maximum of the peak intensity FWHM vs Temperature FWHM nm Temperature K 10 60 110 160 210 260 310 Figure 5 7 FWHM result of the sample no 2 2506 3 68 However the principal data result of PL measurement is the IQE In theory the intensity decreases linearly with temperature But in practice it is not perfectly linear due to temperature control issues In fact the sample is attached to a sample holder in the cold finger The sample holder is equipped with a temperature sensor GaAs diode and a heater which are the fundamentals elements for temperature control and monitoring Therefore the temperature of the sample holder is quite well controlled and known but the thermal contact between the sample holder and the actual sample is done using a thin layer of Apiezon N grease which improves the contact compared to copper tape but its efficiency is nonetheless not accurately known Thus a difference exists between the displayed temperature and the actual sample temperature Therefore a linear regression curve fitting is used on the raw data in order to correct t
82. tion effects are composed of two components piezoelectricity and spontaneous polarization Piezoelectricity is produced by mechanical strain in the crystal the polarization being proportional to the strain and changing sign with it This is also known as the direct piezoelectric effect the converse effect is when a crystal is strained when an electric field is applied In the case of GaN based LEDs the direct effect induces a built in potential because of the strain between different materials due to lattice parameter differences which implies a distortion of the electronic band structure and leads to a spatial separation of the electron and hole wave functions inside the QWs 31 Therefore electrons and holes are separated towards opposite sides of the layer resulting in a reduction in the energy of confined electron hole pairs and a reduced wave function overlap This is referred as the Quantum Confined Stark Effect QCSE 26 The following figure illustrates the band diagram bending in the case of a QCSE Conduction band energy Valence band energy Figure 3 13 Spatial separation of electron and hole wave functions in QCSE 26 3 9 Improvement of LEDs 3 9 1 Hole Transport The study of hole transport is focused on the transport of the holes into the active region when using InGaN instead of GaN p type layer in our LED structure A p GaN layer is commonly grown in blue LEDs because of a better lattice match with the GaN qua
83. tive recombination in the QW but not in the quantum barriers A detector composed of a monochromator and an array of CCDs scans the spectrum and records the intensity at each wavelength Since the measurements are carried out at 300K and the material is not perfect each peak will be broadened and thus have a 42 measurable line width Sub bandgap levels created by dopants and defects can result in photon emission which will broaden the spectrum or produce peaks of their own In the AMDG laboratories PL measurements are performed by two different instruments The first is an Accent RPM 2000 that provides a pump laser beam power of 2 mW ata wavelength of emission at 266 nm The second is a more recent system that I have setup and which I have been assigned to supervise the measurement It is more efficient for blue and green LED applications because the pump laser has an output power that is higher 20 mW and emits at 325 nm which is closer to the target wavelengths so it doesn t enable as much non desired radiative recombination outside of QWs in the active region The system 1s composed of a Helium Cadmium laser used as the optical source a cryostat that cools down the LED sample to 10 K for internal quantum efficiency assessment purpose refer to Chapter 5 a focus lens setup a triple axis grating monochromator and a CCD detector cooled down to 130 C by liquid nitrogen 4 3 Atomic Force Microscopy Atomic force microscopy
84. to the three QWs has been observed The excitation source or optical pump is a Helium Cadmium HeCd Series 74 laser produced by Melles Griot Figure 5 2 Picture of a HeCd Series 74 laser Courtesy to Dong A University The principle of functioning of helium cadmium lasers is a complex electrical discharge that produces output in the ultraviolet 325 nm and a power of 25 mW The basic medium of the discharge is helium a gas but the lasing component is cadmium a metal To obtain stable laser action the cadmium must first sublimate into a quasi gaseous form and then be evenly distributed throughout the helium discharge through the processes of diffusion and cataphoresis 5 3 Cryostat The cryostat system is used to control and monitor the sample temperature It 1s composed of a helium compressor cold head cold finger temperature sensor heater and temperature controller The sample can be cooled down to 10 K in approximately 60 40 minutes The advantage of this system is that helium gas flows in a closed loop which means there is no need of recharging the helium reservoir The compressor is a single stage water cooled rotary compressor designed to deliver high pressure oil free helium gas to the cold head The optimal pressure is around 260 psi and the gas is 99 995 pure The purpose of the compressor is to cool the gas through three circuit heat exchangers remove oil and moisture and pressurize compress the helium
85. ttie and Landsberg 31 3 2 442M Eg T Eg T o tem ET 3 3 kT Where k is the Boltzmann constant T the temperature E T temperature dependent energy bandgap and M the electron to hole mass ration The carrier lifetime expresses the average time during which the effect is occurring A small carrier lifetime means that the effect should be predominant as it is occurring at a faster pace which means that more carriers are impacted by this effect compared to higher lifetime ones where the occurrence is slower 28 As we can see from the equation 3 3 the Auger recombination carrier lifetime is dependent upon the value of the energy gap Therefore a large bandgap energy such as GaN around 3 4 eV has a higher Auger carrier lifetime compared to GaAs about 1 4 eV but the quantum efficiency of GaAs LEDs are quite good which tends to discredit this effect Furthermore theoretical studies have found values of the Auger recombination coefficient for GaN devices which are too small to account for the experimental results 32 However Auger recombination is actually divided in two mechanisms direct and indirect The following figure illustrates the difference between direct and indirect Auger recombination E E Figure 3 12 Illustration of a direct and b indirect Auger recombination 33 The previous discussion was about the direct Auger recombination mechanism The indirect Auger recombination IAR is
86. uminance and power are currently not used and not accurately calculated as they require special setup to be measured such as integrated sphere but they are still of interest for comparison purpose only The variance is pretty high compared to other data because they are directly related to intensity but lower than 3 for the luminance and 4 6 for the power 84 Concerning the saturation this is an indicator of how close the light intensity is to the CCD detector saturation Its variance is inferior to 0 4 The difference that we experience with QuickTest 2 0 doesn t come from the program itself as the core is the same but from the way of the measurement is done Previously the control of the current source meter and the execution of the measurement were done manually which vary from user to user and even from the same user Now all the process is executed automatically which means that no variation is experienced between measurements and leads to a better accuracy This explains why differences in intensity and processed data results exist Therefore the comparison between LED samples is now more efficient than previously which enables the group to get more accurate data 85 7 CONCLUSION The new challenge the world is facing today is to make energy consumption more efficient as the global energy demand is skyrocketing The LED device is one of the most promising technologies that will enable to start the sma
87. weep Current Tab and right List Tab 79 The Sweep Current tab enables to use the source meter in a current sweep mode with a fix increment The starting current is entered at the Start Current control the ending current at the Stop Current and the Actual Current indicates the current that is currently outputted during measurement Concerning the List tab it gives the possibility to enter manually the current values 6 3 5 DC Pulse Mode parameters Tabs DC Mode Pulse Mode DC Mode Pulse Mode Trigger Count Duty Cycle ON time s 5 9 39 0 1 3 s offset Pulse Width ms OFF time s 5 0 15 5 ms 0 Pulse Delay s 0 05 0 9999 999 5 Figure 6 6 left DC Mode Tab and right Pulse Mode Tab On DC Mode the ON time is a delay that increases the time during which the source meter output is ON The minimum is equal to 1 3 s this lead time is due to the electronics response in the source meter The OF F time is the time between two measurements when the source meter output is shut off for a very short moment range of ms This control allows entering a delay to increase the shut off time In Pulse Mode drive current is modulated vs time using a step function the pulse configuration is shown in the following figure 80 Pulse measure timing for default source measure configuration Meas lt Delay oe 80us a Sig 4 Pulse Width Output On Time Output Off Time OV or OA Meas lt 2 9
88. y 13 885x 485 89 R 0 9979 800 Rsh 2 18 kQ sq 00 pS NA Re 242 5 Q Figure 4 6 Example of TLM result AMDG Internal Document 48 4 5 Hall Effect measurement The Hall effect is the generation of a voltage difference the Hall voltage across an electrical conductor transverse to an electric current in the conductor and a magnetic field perpendicular to the current It was discovered by Edwin Hall in 1879 Figure 4 7 Illustration of Hall effect in a p type bar 38 If a magnetic field is applied perpendicular to the direction in which carriers drift in a semiconductor the path of the carriers tends to be deflected The total force F Newton on a single carrier due to the electric and magnetic fields 1s F q E vxB 4 1 Where q is the magnitude of the electronic charge Coulomb v the drift velocity of the carrier m s B the magnetic field vector Tesla and E the electric field N C 49 In the y direction the force is F q Ey v B 4 2 Thus a force represented by qgv B 1s experienced in the y direction To maintain a steady state flow of carriers holes in this example down the length of the bar the electric field E must balance the product v B By vB 4 3 E J4q Po B 4 4 After some calculations the hole concentration 1s ly Bz 4 5 Po q t Var 4 5 With po the hole concentration cm I the electrical current A t the thickness o

a study of efficiency droop of green light emitting

Contents

Download Pdf Manuals

Related Search

Related Contents