Diffusion injected light emitting diode - Aaltodoc - Aalto
advertisement
D e p a r t me nto fM i c r o -a ndN a no s c i e nc e s L auri Riut t ane n D if f usio n inje c t e dl igh te mit t ing dio de A a l t oU ni v e r s i t y 2 0 1 5 D iffusio n inje c t e dl igh t e mit t ing dio de L a ur iR i ut t a ne n D O C T O R A L D I S S E R T A T I O N S Aalto University publication series DOCTORAL DISSERTATIONS 213/2015 Diffusion injected light emitting diode Lauri Riuttanen A doctoral dissertation completed for the degree of Doctor of Science (Technology) to be defended, with the permission of the Aalto University School of Electrical Engineering, at a public examination held at the large seminar room of the Micronova-building on 18 December 2015 at 12. Aalto University School of Electrical Engineering Department of Micro- and Nanosciences Supervising professor Professor Markku Sopanen, Aalto University, Finland Thesis advisor Doctor Sami Suihkonen, Aalto University, Finland Preliminary examiners Doctor Rachel Oliver, University of Cambridge, United Kingdom Professor Magnus Borgström, Lund University, Sweden Opponent Professor Charles Thomas Foxon, University of Nottingham, United Kingdom Aalto University publication series DOCTORAL DISSERTATIONS 213/2015 © Lauri Riuttanen ISBN 978-952-60-6581-6 (printed) ISBN 978-952-60-6582-3 (pdf) ISSN-L 1799-4934 ISSN 1799-4934 (printed) ISSN 1799-4942 (pdf) http://urn.fi/URN:ISBN:978-952-60-6582-3 Unigrafia Oy Helsinki 2015 Finland Abstract Aalto University, P.O. Box 11000, FI-00076 Aalto www.aalto.fi Author Lauri Riuttanen Name of the doctoral dissertation Diffusion injected light emitting diode Publisher School of Electrical Engineering Unit Department of Micro- and Nanosciences Series Aalto University publication series DOCTORAL DISSERTATIONS 213/2015 Field of research Optoelectronics Manuscript submitted 10 August 2015 Date of the defence 18 December 2015 Permission to publish granted (date) 17 November 2015 Language English Monograph Article dissertation (summary + original articles) Abstract Lighting plays a major role in consumption of electrical energy in the world. Thus, increasing the efficiency of light sources is one key element in reducing the green house gas emissions. Light emitting diodes (LEDs) are gaining a foothold in general lighting. Despite their rapid development in light output and their superior efficiency compared to other light sources, LEDs still need improvements in order to become the ultimate lighting technology. A typical LED is a double heterojunction (DHJ) structure, in which the active region fabricated from a lower band gap material is sandwiched between higher band gap p- and n-doped regions. By biasing such a structure electrons and holes are transferred by current into the active region, where they recombine releasing energy as photons. The carrier injection in a conventional LED structure is typically efficient. However, in more exotic novel structures based on nanowires or near surface nanostructures, fabricating a DHJ becomes difficult. This thesis presents the experimental studies on a novel current injection method for light emitting applications. The method is based on bipolar diffusion of charge carriers. Unlike in the conventional method, the active region does not have to placed between the p- and n-layers of the pn-junction. The diffusion injection method is experimentally demonstrated by two types of prototype structures. The first prototype was fabricated using a multi quantum well (MQW) stack buried under the pn-junction. The second prototype was fabricated using a near surface quantum well (QW) placed on top of the pn-junction. The first prototype showed that the diffusion current components can be used to excite an active region outside of the pn-junction. The second prototype showed a large improvement in injection efficiency as well as the suitability of the method for exciting surface structures. The applications of diffusion injection can be found in blue galliun nitride based LEDs studied in this thesis as well as in green solid-state light sources, light sources integrated into silicon technology and devices based on nanostructures and plasmonics. Keywords gallium nitride, III-nitrides, LED, diffusion ISBN (printed) 978-952-60-6581-6 ISBN (pdf) 978-952-60-6582-3 ISSN-L 1799-4934 Location of publisher Helsinki Pages 121 ISSN (printed) 1799-4934 Location of printing Espoo ISSN (pdf) 1799-4942 Year 2015 urn http://urn.fi/URN:ISBN:978-952-60-6582-3 Tiivistelmä Aalto-yliopisto, PL 11000, 00076 Aalto www.aalto.fi Tekijä Lauri Riuttanen Väitöskirjan nimi Diffusiolla viritetty loistediodi Julkaisija Sähkötekniikan korkeakoulu Yksikkö Mikro- ja nanotekniikan laitos Sarja Aalto University publication series DOCTORAL DISSERTATIONS 213/2015 Tutkimusala Optoelektroniikka Käsikirjoituksen pvm 10.08.2015 Julkaisuluvan myöntämispäivä 17.11.2015 Monografia Väitöspäivä 18.12.2015 Kieli Englanti Yhdistelmäväitöskirja (yhteenveto-osa + erillisartikkelit) Tiivistelmä Valaistus kuluttaa merkittävän osan sähköenergiasta maailmanlaajuisesti. Valaistuksen hyötysuhteen parantaminen on siis yksi merkittävimmistä menetelmistä vähentää kasvihuonepäästöjä. Vaikka loistediodien (LEDs, engl. light emitting diodes) nopea kehitys ja niiden ylivoimainen hyötysuhde kilpaileviin teknologioihin on antanut LED:eille jalansijaa myös yleisvalaistuksessa, LED-teknologia tarvitsee vielä tutkimusta, jotta ne voisivat täydellisesti korvata vaihtoehtoiset teknologiat. Perinteinen LED perustuu kaksoisheteroliitokseen (DHJ, engl. double heterojunction), jossa aktiivinen alue on valmistettu p- ja n-tyyppiseksi seostettujen alueiden väliin. Kytkemällä jännite rakenteen yli elektronit ja aukot liikkuvat sähkökentän mukaisesti ajautumisvirtana seostetutuista alueista kohti aktiivista aluetta, jossa ne rekombinoituvat vapauttaen energiaa fotoneina. Varaustenkuljettajien syöttö on perinteisessä LED-rakenteessa tyypillisesti erittäin hyvä. Haasteita ilmenee uusissa, eksoottisissa LED-rakenteissa, jotka perustuvat lähellä pintaa oleviin nanorakenteisiin. Näissä tapauksissa DHJ:n valmistaminen on hankalaa ja varaustenkuljettajien syöttö on usein heikkoa. Tämä väitöskirja esittelee kokeellisia tutkimuksia uudesta diffuusioon perustuvasta LEDrakenteiden viritysmenetelmästä. Menetelmässä aktiivisen alueen ei tarvitse olla seostettujen alueiden välissä. Kokeelliset todisteet menetelmän toimivuudesta näytetään kahdella eri tyyppisellä prototyyppirakenteella. Ensimmäinen prototyyppi on pn-liitoksen alle haudattu monikvanttikaivorakenne. Toisessa prototyypissä yksittäinen kvanttikaivo on pn-liitoksen päällä. Ensimmäinen koerakenne osoitti, että pn-liitoksen ulkopuolelle valmistetun aktiivisen alueen viritys onnistuu. Toisella rakenteella havaittiin merkittävä hyötysuhteen parannus. Se myös osoitti, että pinnassa sijaitseva rakenne voidaan virittää sen alla olevalla pn-liitoksella. Mahdollisia sovelluksia menetelmälle löytyy jo mainittujen sovellusten lisäksi vihreiden puolijohdevalolähteiden valmituksessa sekä pii-teknologiaan integroitavissa valolähteissä. Avainsanat galliunnitridi, III-nitridit, LED, diffuusio ISBN (painettu) 978-952-60-6581-6 ISBN (pdf) 978-952-60-6582-3 ISSN-L 1799-4934 Julkaisupaikka Helsinki ISSN (painettu) 1799-4934 Painopaikka Espoo ISSN (pdf) 1799-4942 Vuosi 2015 Sivumäärä 121 urn http://urn.fi/URN:ISBN:978-952-60-6582-3 Preface The work presented in this thesis was carrier out in Department of Microand Nanosciences in School of Electrical Engineering of Aalto University. Firstly, I want to thank professor Markku Sopanen for giving me this opportunity to work, study and perform my research in his group under his guidance. Secondly, I want to express my gratitude for Dr. Sami Suihkonen for instructing me in my research as well as for his insight and expertise in the field. I would also like to thank my colleagues, who have become more than co-workers but also good friends. In addition to their work-related discussions, the good company and off-topic discussions have made my research more than just work but also a pastime. Especially, I would like to thank Dr. Henri Nykänen for his expertise in processing and cleanroom work, Dr. Pyry Kivisaari and Dr. Jani Oksanen for their simulation work and helping me to understand the physics of LEDs, and Olli Svensk for his know-how in MOVPE growth. My gratitudes also go for the University of Jyväskylä, whose National Doctoral Programme in Nanoscience (NGS-Nano) funded a significant part of my research. My friends and family have been the most supportive during my time in studies and my life in general. My final gratitudes goes to my lovely wife Sini for her love and support. He, who controls ammonia, controls MOVPE. Ammonia must flow. Espoo, November 30, 2015, Lauri Riuttanen i Preface ii Contents Preface i Contents iii List of Publications v Author’s Contribution vii List of abbreviations ix List of symbols xi 1. Introduction 1 2. Background 5 2.1 History of LEDs . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2.2 History of III-nitrides . . . . . . . . . . . . . . . . . . . . . . . 6 3. Electrical properties of semiconductors 9 3.1 Band theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 3.2 Current transport in semiconductors . . . . . . . . . . . . . . 10 3.3 Polarisation fields and piezoelectricity in semiconductors . . 13 3.4 Carrier lifetime in semiconductors . . . . . . . . . . . . . . . 14 3.5 Basic operating principle of a light emitting diode . . . . . . 16 4. III-nitride light emitting diodes 21 4.1 Properties of III-nitrides . . . . . . . . . . . . . . . . . . . . . 21 4.2 Efficiency droop in III-nitride light emitting diodes 24 . . . . . 5. Experimental 27 5.1 Metal organic vapour phase epitaxy . . . . . . . . . . . . . . 27 5.2 Growth of III-nitride light emitting diode . . . . . . . . . . . 28 iii Contents 5.3 Measuring carrier lifetimes . . . . . . . . . . . . . . . . . . . 30 5.4 Measuring droop . . . . . . . . . . . . . . . . . . . . . . . . . . 32 6. Diffusion injected light emitting diode 35 6.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 6.2 Current Transport in DILED . . . . . . . . . . . . . . . . . . 36 6.3 Buried MQW DILED . . . . . . . . . . . . . . . . . . . . . . . 36 6.4 Surface QW DILED . . . . . . . . . . . . . . . . . . . . . . . . 43 6.5 Future prospects of DILED structures . . . . . . . . . . . . . 45 7. Conclusion 49 References 51 Errata 57 Publications 59 iv List of Publications This thesis consists of an overview and of the following publications which are referred to in the text by their Roman numerals. I L. Riuttanen, P. Kivisaari, N. Mäntyoja, J. Oksanen, M. Ali, S. Suihkonen, and M. Sopanen. Recombination lifetime in InGaN/GaN based light emitting diodes at low current densities by differential carrier lifetime analysis. Physica Status Solidi C, 10, 327-331, January 2013. II P. Kivisaari, L. Riuttanen, J. Oksanen, S. Suihkonen, M. Ali, H. Lipsanen and J. Tulkki. Electrical measurement of internal quantum efficiency and extraction efficiency of III-N light-emitting diodes. Applied Physics Letters, 101, 021113, July 2012. III L. Riuttanen, P. Kivisaari, H. Nykänen, O. Svensk, S. Suihkonen, J. Oksanen, J. Tulkki, and M. Sopanen. Diffusion injected multi-quantum well light-emitting diode structure. Applied Physics Letters, 104, 0811002, February 2014. IV L. Riuttanen, P. Kivisaari, O. Svensk, J. Oksanen and S. Suihkonen. Diffusion Injection in a Buried Multiquantum Well Light-Emitting Diode Structure. IEEE Transactions on Electron Devices, 62, 902-908, March 2015. V L. Riuttanen, P. Kivisaari, O. Svensk, T. Vasara, P. Myllys, J. Oksanen and S. Suihkonen. Vertical excitation profile in diffusion injected multiquantum well light emitting diode structure. Proc. SPIE, 9363, 93632A, v List of Publications March 2015. VI L. Riuttanen, P. Kivisaari, O. Svensk, J. Oksanen and S. Suihkonen. Electrical injection to contactless near-surface InGaN quantum well. Applied Physics Letters, 107, 051106, August 2015. vi Author’s Contribution Publication I: “Recombination lifetime in InGaN/GaN based light emitting diodes at low current densities by differential carrier lifetime analysis” The author wrote the manuscript and designed the measurement setup. The measurements and data handling was performed by the author. Publication II: “Electrical measurement of internal quantum efficiency and extraction efficiency of III-N light-emitting diodes” The author designed the measurement setup and performed the measurements. Publication III: “Diffusion injected multi-quantum well light-emitting diode structure” The author wrote the manuscript. Growth and processing of the studied samples as well as the measurements were performed by the author. Publication IV: “Diffusion Injection in a Buried Multiquantum Well Light-Emitting Diode Structure” The author wrote the manuscript. Growth and processing of the studied samples as well as the measurements were performed by the author. vii Author’s Contribution Publication V: “Vertical excitation profile in diffusion injected multi-quantum well light emitting diode structure” The author wrote the manuscript. Growth and processing of the studied samples as well as the measurements were performed by the author. Publication VI: “Electrical injection to contactless near-surface InGaN quantum well” The author wrote the manuscript. Growth and processing of the studied samples as well as the measurements were performed by the author. viii List of abbreviations AC Alternating current AlN Aluminium nitride AlGaN Aluminium gallium nitride BN Boron nitride CCS Close coupled shower head CTE Coefficient of thermal expansion DADR Density activated defect recombination DC Direct current DHJ Double heterojunction DILED Diffusion injected light emitting diode EBL Electron blocking layer EL Electroluminescence EQE External quantum efficiency EXE Extraction efficiency GaAs Gallium arsenide GaAsP Gallium arsenide phosphide GaN Gallium nitride GE General Electric HVPE Hydride vapour phase epitaxy ICP-RIE Inductively coupled plasma-reactive ion etching InN Indium nitride InGaN Indium gallium nitride IQE Internal quantum efficiency I-V Current-voltage LD Laser diode LED Light emitting diode LEEBI Low energy electron beam irradiation MBE Molecular beam epitaxy ix List of abbreviations x MIS Metal-insulator-semiconductor MOVPE Metal organic vapour phase epitaxy MOCVD Metal organic chemical vapour deposition MQW Multi quantum well OLED Organic light emitting diode PL Photoluminescence QCSE Quantum confined Stark effect QD Quantum dot QW Quantum well RCA Radio Corporation of America SiC Silicon carbide SL Superlattice SQW Single quantum well SRH Shockley-Read-Hall TMA Trimethylaluminium TMG Trimethylgallium TMI Trimethylindium TRPL Time resolved photoluminescence UV Ultraviolet WPE Wall plug efficiency List of symbols A Shockley-Read-Hall recombination coefficient Apn Cross sectional area of a pn-junction B Radiative recombination coefficient c Speed of light C Auger recombination coefficient CLED Capacitance of light emitting diode d Thickness of active region D2 Second derivative Dn Diffusion constant for electrons E Electric field EC Conduction band energy EF Fermi-level energy EF,n Electron quasi-Fermi-level EF,p Hole quasi-Fermi-level Eph Photon energy EV Valence band energy f Frequency h Plank constant I Current J Current density kB Boltzmann coefficient L Distance Ls Series inductance n Carrier density nn Electron density np Hole density Pm Maximum of the optical power PY B Optical power emitted at yellow band xi List of symbols xii PQW Optical power emitted by quantum well q Elementary charge R Recombination rate Rd Differential resistance Rs Series resistance T Temperature V Voltage Va Applied electric potential Vi Inner electric potential ηinj Injection efficiency ηm Maximum External quantum efficiency λ Wavelength μ Electron mobility τ Carrier lifetime φ Phase difference χ External extraction efficiency ∇ Gradient operator 1. Introduction Light emitting diodes (LEDs) have unquestionable impact on our everyday lives. LEDs can be found in almost every electronic device in hand. They are nowadays not only used as indicator lights as they were in the early years of long history of LEDs, but they have gained ground also in solid state lighting and screen back lighting of hand held devices and televisions. The first report of electroluminescence from semiconductors was hundred years ago, and nowadays the LED industry has grown in to a billion dollar business. The hard work and key inventions in developing blue LEDs, which are used in fabricating white light emitters, led The Royal Swedish Academy of Sciences to award Nobel price in physics to Isamu Akasaki, Hiroshi Amano and Shuji Nakamura in 2014. There is demand for more and more efficient light sources, since lighting is the largest single user of electric energy and a rapidly growing source of energy demand and greenhouse gas emissions. In 2005 the electricity consumed by lighting was 2650 TWh worldwide, about 19 % of the total global electricity consumption [1]. The incandescent light bulb is becoming history in lighting due to its poor efficiency and they are slowly being replaced by more efficient and longer lifetime light sources such as fluorescent tubes and LEDs. Even though LEDs have become a mature technology in growth and device engineering point of views, there is still room for improvement and research. Almost whole visible spectrum can be nowadays covered by LEDs. However, the efficiency of green LEDs is reasonably poor [2]. This is often referred to as the green gap. Moreover, LEDs, especially gallium nitride (GaN) based white light emitters, suffer from a phenomenon called the efficiency droop [3]. The efficiency of an LED is dependent on the injection current density. At low injection current densities the LEDs operate at modest efficiency. When injection current density is increased, the ef- 1 Introduction ficiency improves. When the peak efficiency is achieved, the efficiency starts to decrease. This decrease or roll over of efficiency is the droop effect. Several, some even contradictory, theories have been presented in the research community, but no overall agreement has been gained. Solving the efficiency droop is one key element in order to achieve large scale deployment of LED based high power light sources for solid state lighting and automobile headlights. In order to improve the efficiency of LEDs and also to avoid the droop, designing novel LED structures has received an increasing amount of attention in the research community. Since the efficiency droop is dependent on the carrier density in the active region, increasing the volume of the light emitting region of the device has been suggested as a solution. The typical light emitting region in GaN based LEDs is quantum well (QW) based structure. A QW is a thin film of lower band gap material sandwiched in between of higher band gap materials called barriers. The thickness of the well is only a few nanometers and the thickness of the barriers is roughly at least 10 nm. The QW structure confines charge carriers allowing an efficient light generation. Growing thicker QWs increases the volume, but results in decreased radiative recombination due to electron and hole separation in the QW. Increasing the amount of QWs results in poor electron and hole distribution along the multi quantum well (MQW) stack. Nanowires are considered to have potential to achieve the ultimate efficiency of LEDs. Their superior crystal quality reduces significantly the unwanted nonradiative recombination in the device structure. The carrier confinement is also improved compared to thin film devices. In addition, light extraction from the device itself is significantly higher than in the conventional planar LED design. Similar arguments have also been used for fabricating quantum dot (QD) LEDs. Since the LEDs are fabricated from materials, which have higher refractive index, part of the generated light is trapped inside the device structure. This reduces light extraction properties of the device. For improving light extraction from planar LEDs, structures exploiting surface plasmons have been developed. Practically all modern inorganic LED designs are based on a double heterojunction (DHJ) structure, which is in practice a sandwich-like structure. The light emitting region is fabricated in the middle of the device structure. These structures and the associated current injection principle have remained essentially unchanged since the invention of the pnjunction based LED. The DHJ structure works well in the conventional 2 Introduction approach, in which the LED is planar and the light emitting region is a stack of thin films. However, the conventional DHJ design does not adapt well to current injection of nanostructures or active regions located close to surfaces. In such cases challenges arise in, e.g., the fabrication of the nanostructured DHJs [4], the formation of depletion layer and the resulting large leakage currents [5]. This hinders the development of emerging optoelectronic devices, such as photonic crystal [6], surface plasmon enhanced [5, 7–11], nanowire [4, 12, 13] and quantum dot [14] structures. This thesis mainly discusses the research and the studies of fundamentally different approaches for electrically excite light emitting materials. In a conventional LED structure diffusion of minority charge carriers may extend over relatively long distances even in the presence of the DHJ potential barriers designed to reduce the detrimental leakage currents [15, 16]. The studied approach is based on utilizing the mentioned diffusion currents. This allows the light emitting region to be left outside the pn-junction and, therefore, it can be fabricated completely free of contact structures. Additionally, if the active region is buried under the pnjunction, it can be left completely unetched. The diffusion injected light emitting diode (DILED) structures studied in this work are modifiable to surface plasmon utilizing, QD, nanowire and large area LEDs as well as solar cells and detectors. The efficiency of the first prototype DILED structure was in the same range than that of the reported first GaN based DHJ LEDs. The efficiency of a second generation prototype was already in the range of that of a simple reference LED fabricated with similar active region. However, the enormous potential in the DILED structure in terms of device engineering and utilization of huge variety of different materials opens up a whole new world for LEDs. The background for this thesis including the history of LEDs and IIInitrides is handed out in chapter 2. Introduction into electronic properties of semiconductors is given in chapter 3. III-nitrides in general and III-nitride LEDs are discussed in chapter 4. Chapter 5 covers the experimental section. In chapter 6 the fabricated DILED structures are presented and their properties are thoroughly discussed as well as the future of DILEDs. Finally the thesis is concluded in chapter 7. 3 Introduction 4 2. Background 2.1 History of LEDs In the beginning of the 20th century, H. J. Round was studying rectifying properties of a silicon carbide (SiC) and metal junction as he noted that the sample was emitting light under bias voltage. In 1907 he wrote a paper, in which he described the light emitting effect in a metal-semiconductor junction. This was the first study, which reported electroluminescence. However, since semiconductors in that time were not well known, he reckoned that the light was generated by thermoelectric effect in the junction. It required few decades until the phenomenon was explained by modern semiconductor theories. [17] Lossev, in 1928, reported his investigations on this mystical luminescence properties of SiC based rectifiers. He noticed, that the emission was not based on incandescence, but it is linked into some electrical phenomenon. [18] However, the phenomenon did not receive large scale interest until 1940’s, when the pn-junction was invented by R. Ohl. [19] R. Ohl noticed that the impurities in silicon samples, formed by the fabrication processes, created regions which affected the conductivity of the silicon. Current transport in some of these regions was dominated by carriers with negative charge and in part it was dominated by carriers with positive charge. These regions were named as n-type and p-type material, respectively. [19] When sample consisted of both p- and n-type material, the sample exhibited current rectifying properties. However, the theory of pn-junction was not formed until 1947, when J. Bardeen and W. Brattain invented the transistor in Bell Laboratory. This was followed by the W. Shockley’s theoretical studies on the pn-junction in 1948. [19] A huge step towards modern LEDs occurred in 1950’s, when the era 5 Background of III-V semiconductor family started. III-V semiconductors were prior completely unknown material family, since they do not occur in nature. The novel materials were proven to be optically very active. In 1954 bulk growth of gallium arsenide (GaAs) was initiated and it was used for epitaxial substrates for pn-junction diode structures. The first commercial LED was fabricated by Texas Instruments in the beginning of 1960’s. It was gallium arsenide (GaAs) based device emitting at 870 nm. However, large manufacturing costs kept the production scale small. [20] Holonyak and Bavacqua reported in 1962 a coherent light emission from gallium arsenide phosphide (GaAsP) based laser diode (LD) operating at low temperatures. The device functioned as an LED at room temperature. This report can be seen as the beginning of the conquest of LEDs in visible range. The first commercial LED emitting at visible range, based on GaAsP, was manufactured by General Electric (GE) Corporation in the early 1960’s. The cost of a single LED was significantly high. Mass production of LEDs began in 1968 by Monstanto Corporation. This initiated the era of solid state lamps. [20] 2.2 History of III-nitrides The history of III-nitrides is long, even though the blue LED, which is currently the main application for III-nitrides, is a reasonably recent invention. In year the 1938 Juza and Hahn managed to fabricate small spikes of GaN by flowing hot ammonia over molten gallium. Grimmeiss, et al., used the same method and achieved small GaN crystals. [21] At the end of the 1960’s James Tietjen, who at the time was head of the material research group at Radio Corporation of America (RCA), wanted to develop LED based television, which could be hanged on the wall like a painting. For achieving complete colour map, three main colours are needed: red, green and blue. The red LED was obtainable by using the GaAsP alloy and green from GaP. Therefore, only missing piece was the bright blue LED. [20] Tietjen challenged Paul Maruska and his research team for searching a way to fabricate single crystal GaN films, with which Tietjen believed to achieve the blue LED. Maruska had earlier fabricated GaAsP based red LEDs and had a lot of experience from the advantages and disadvantages of III-V semiconductors. Maruska began his work by studying research articles from 1930’s and 1940’s, which describe the mentioned 6 Background molten gallium and ammonia method. He chose sapphire as a substrate, since sapphire does not react easily with ammonia. [20] The first breakthrough occurred in 1969, when Maruska and Tietjen succeeded to grow a single crystal GaN film on a sapphire substrate by using hydride vapour phase epitaxy (HVPE) [22]. Maruska noticed that the grown non-doped GaN films had naturally n-type conducting properties without any doping. They started to find a dopant for achieving p-type conductivity, but failed. In the summer of 1971 Pankove from RCA demonstrated the first electrically excited light emitter fabricated from GaN. This metal-insulator-semiconductor (MIS) diode emitting blue and green light was fabricated from unintentionally doped GaN, insulating zinc doped GaN and indium. In the same year Dingle, et al., demonstrated optically pumped ultraviolet (UV) laser operating at 2 K. In the year 1974 Tietjen ended his research, after which GaN research was almost completely dead for decades. Exception was Japanese Isamu Akasaki, who stubbornly continued GaN research with his colleagues. [20] In the beginning of 1970’s, Akasaki decided to focus on GaN based blue emitter, which would utilize pn-junction [22]. GaN was only grown using HVPE, until Akasaki succeeded to grow GaN with molecular beam epitaxy (MBE) in 1974. Even though the quality of the grown film was reasonably non-uniform, the breakthrough yielded him three year funding for developing GaN based light emitters. In the year 1978 Akasaki’s research group fabricated a blue LED with an efficiency of 0.12 %, which was higher than ever before. No significant improvement in GaN research occurred until mid-1980’s. [23] Akasaki understood the importance of choosing the correct growth method. Akasaki and his research group began working on the GaN growth on sapphire with metal organic vapour phase epitaxy (MOVPE) at the University of Nagoya. By 1980 GaN could be grown with MBE, HVPE and MOVPE. The growth rate with HVPE is too high for controllable thin film growth. MBE during that time was too slow and GaN during MBE growth was prone to nitrogen desorption. The GaN growth rate with MOVPE was intermediate between the growth rates of HVPE and MBE. Akasaki chose sapphire as the substrate material. [23] The results were not particularly good, which Akasaki believed to be due to differences of coefficient of thermal expansion (CTE) and lattice constant between GaN and sapphire. Akasaki, et al., tried to mimic homoepitaxy by using a low temperature grown buffer layer before growing the final layer structure. This reduced 7 Background the stress in the grown films significantly. In 1985 they were the first, who succeeded to grow high quality GaN films. [22] Several research groups tried to grow p-type GaN without any success. Most research groups gave in, while Akasaki et al. refused to quit. [20] In 1989 Amano et al. first demonstrated p-type conductivity in GaN [24]. The key factor was to notice that the acceptors could be activated by using low energy electron beam irradiation (LEEBI) on Mg doped GaN. In the year 1990 Shuji Nakamura developed a new two flow MOVPE, which improved the crystal quality significantly [25]. These events led to the first blue LED based on the GaN pn-junction demonstrated in 1991 by Nakamura’s employer Nichia. [26] In 1992 Nakamura et al. noticed that the acceptors in Mg doped GaN could also be activated by thermal annealing instead of LEEBI [27]. The success in p-type doping opened the gate for efficient GaN LEDs and also for laser diodes (LDs). [20] In 1996 Nakamura demonstrated the first continuouswave GaN based LD operating at room temperature [26]. GaN based light emitting diodes have been under extensive studies and the technology is becoming mature. The external quantum efficiency (EQE) of white LEDs already exceeds 80 % [28]. However, there are yet challenges to overcome. InGaN LEDs suffer from a phenomenon called the efficiency droop, which is reduction of quantum efficiency at high injection currents [3]. More discussion on the droop effect can be found in sections 4.2 and 5.4. Efficiency of LEDs could be enhanced significantly by utilizing, e.g., nanowires [4, 12, 13], quantum dots [14] or plasmonic gratings [5, 7–11]. However, these structures face challenges in current injection. The diffusion injection discussed in this thesis can be used for circumventing the current injection related challenges. In addition, semiconductor LEDs face a challenge, called "the green gap", which refers to the lack of high efficiency LEDs in the green spectral range [2]. Utilizing nanowires, the indium composition in InGaN can be extended to higher compositions allowing the emission peak to reach the green and even the red spectral range [29]. Diffusion injection can, therefore, also help closing "the green gap" by green light emitting nanowire LEDs. 8 3. Electrical properties of semiconductors Semiconductors are materials with conductivity somewhere between insulators and conductors. Semiconductors can be conducting or insulating depending on the circumstances. Current transport in semiconductors can be explained through band theory discussed in the following section. 3.1 Band theory The electrical, and partly also the optical properties of semiconductors are often described through band theory. If one atom is considered, it has defined energy states which electrons can occupy. If two atoms are brought together, the Pauli exclusion principle dictates that the electrons, or fermions more generally, cannot occupy the exact same quantum state. Therefore, the electron quantum states have to misalign slightly. The electron states in solid materials having a very large number of atoms can be described as semi-continuous energy bands. Forbidden energy states are left between the bands. At absolute zero, i.e., at 0 K, the highest energy band with occupied states is defined as the valence band. In theory, the valence band of semiconductors is completely filled at absolute zero leaving no free electrons for conduction. At room temperature some of the electrons in the valence band are excited to the energy band higher than the valence band. This energy band is referred to as the conduction band. The excited electrons leave an unoccupied state into the valence band. If the energy band is partially filled, the electrons in the band can move. The movement of electrons causes current in the material. Intrinsic semiconductor material is not typically very conductive. By introducing impurities into the semiconductor, its electrical conductivity can be altered. This is called doping. By introducing donor dopant atoms, which give excess electrons to the lattice of the semiconductor, the elec- 9 Electrical properties of semiconductors trical conductivity of the semiconductor increases due to the free electrons in the conduction band. On the other hand, if the semiconductor is doped with acceptor atoms, which remove electrons from the semiconductor, the removed electrons leave unoccupied states into the valence band of the semiconductor increasing the electrical conductivity. A semiconductor with excess free electrons in the conduction band is called an n-type semiconductor and with unoccupied states in the valence band is called a p-type semiconductor. Doping of semiconductors is the base for most of the semiconductor applications, including LEDs. The region of forbidden energy states between the conduction and the valence band is called the band gap. If radiative recombination occurs, the band gap defines the energy and at the same time the wavelength of the emitted photon. The energy of the valence and the conduction band varies in momentum space or k-space, in which k refers to the wavevector. If the maximum of the valence band and the minimum of the conduction band are located in the same location in k-space, the band gap is referred to as a direct band gap, otherwise as an indirect band gap. Figure 3.1 shows a schematic illustration of a direct and an indirect band gap. The semiconductor with a direct band gap has a higher probability of radiative recombination than with an indirect band gap, since radiative recombination in an indirect band gap material requires a shift in the k-space, which is only possible, if the recombining electron interacts with a phonon. Additionally, the valence band splits in three distinct sub-bands, namely heavy hole, light hole and spin-orbit band. The band splitting affects the material band gap, but also the transport properties of the charge carriers are different in each sub-band. 3.2 Current transport in semiconductors Current transport in semiconductors is due to movement of electrons in the conduction band or in the valence band. Since the valence band is mostly filled with electrons and only partially filled with unoccupied states, the movement of electrons in the valence band can be seen also as movement of the unoccupied states. This leads to the basic concept of the hole, which is simply an unoccupied state in the valence band. A hole has effectively the charge of positive elementary charge, while an electron has the charge of negative elementary charge. Since the movement of electrons and holes in practice transfers charge, they are referred to as 10 Electrical properties of semiconductors Figure 3.1. Schematic illustration of a) direct band gap and b) indirect bad gap. Radiative recombination in indirect band gap material requires change in momentum. The valence band splits in three distinct sub-bands, heavy hole, light hole and Spin-Orbit band. charge carriers. As mentioned, both holes in the valence band and electrons in the conduction band contribute to the electrical conduction. The carriers are transported along the semiconductor as a result of an electric potential difference or by diffusion, referred to as drift and diffusion currents, respectively. The potential difference causes an electric field in the medium. Since both electrons and holes are charged particles, the electric field induces a force to the charge carrier causing movement in the electric field’s direction or against it, depending on the charge polarity of the carrier. Diffusion, in general, is caused by thermal motion. However, unlike random walk, which is also caused by thermal motion, diffusion leads into a net flow towards areas, which have lower concentration instead of zero net flow in the case of random walk. The drift-diffusion equation for net current density J for, e.g., electrons can be written as J = qnn μn E + qDn ∇nn , (3.1) in which q is the elementary charge, nn electron density, μn electron mobility, E electric field, Dn the diffusion constant of electrons and ∇ the gradient operator. The first part of the right hand side is the net drift current and the latter part is the diffusion current. The diffusion constant Dn for electrons is a material specific constant, which has temperature 11 Electrical properties of semiconductors dependence following Einstein’s relation Dn = μn kB T, (3.2) in which kB is the Boltzmann constant and T absolute temperature. A pn-junction is the most commonly used semiconductor structure. It is found in diodes, bipolar transistors and LEDs. It is formed when ndoped material is in contact with p-doped material. When the p- and n-doped materials are separated their Fermi-levels are differing. When they are placed in contact charge carrier diffusion begins transporting carriers from the other side to the other. The p-doped region becomes negatively charged and the n-doped region positively charged. These charges cause an electrostatic potential difference between the n- and p-doped region. This electrostatic potential compensates the difference between the Fermi-levels causing the conduction and the valence band to bend. Figure 3.2 illustrates this schematically. The concentration of charge carriers becomes low in the junction and a depletion region is formed. The width of the depletion region depends on the doping levels of the n- and p-doped regions. (a) n-type (b) p-type EC EF EC EV EF EV EC qVi EF EV Figure 3.2. Band diagrams of (a) n- and p-type semiconductors separately and (b) pnjunction in thermal equilibrium. The conduction band (EC ) and the valence band (EV ) bend as the internal potential (Vi ) compensates the differences in the Fermi-levels (EF ). q is the elementary charge. The electrons and the holes experience the bent bands as a potential barrier hindering their transport across the junction. Under external reverse bias, i.e., the n-doped region connected to the higher potential, the 12 Electrical properties of semiconductors depletion region in the junction widens and the inner potential barrier becomes higher. Under forward bias, i.e., the p-doped region connected to the higher potential, the depletion region shrinks and the inner potential barrier becomes lower. Eventually, when enough external forward bias is applied the conduction and valence bands are levelled and the charge carriers can flow along the externally introduced electric field. Under biased conditions there is no well defined Fermi-level in the junction, but the original Fermi-level is split into two position dependent quasi-Fermilevels, one for each type of charge carriers. Figure 3.3 shows the band diagram of a pn-junction under reverse bias and forward bias. (a) (b) EC q(Vi-Va) EC EF,n qVa EF,p EV q(Vi-Va) EF,n EF,p EV qVa Figure 3.3. Band diagrams of a pn-junction under (a) forward and (b) reverse bias. The applied bias Va splits the Fermi-energies of electrons and holes into corresponding quasi-Fermi-levels EF,n and EF,p , respectively. Under forward bias there are four current components, two for both types of charge carriers. The drift current, i.e., the current flowing along the external electric field, is transporting charge carriers through the n- or p-doped material. The diffusion current is the dominant transport mechanism in the depletion region in the pn-junction. Depletion region similar to pn-junction is also formed near the metal contacts used to apply the external field, and thus the diffusion is the dominant transport mechanism to and from the metal contacts. 3.3 Polarisation fields and piezoelectricity in semiconductors Built-in electric fields called polarisation fields may occur in polar crystalline materials. These polarisation fields can either be induced spontaneously or by strain. Total polarisation in the absence of external electric fields is then the sum of spontaneous and piezoelectric (strain induced) polarisation. To understand the origin of polarisation fields, the concept of electronegativity must be considered. Electronegativity describes the ability of an atom to attract electrons toward itself. Electronegativity depends on the atomic number and the electron configuration of the element, thus every 13 Electrical properties of semiconductors element has different electronegativity. When two different atoms form a bond between each other, the electron cloud is generally biased towards the atom, which is more electronegative. Therefore, the bonds between atoms can have permanent dipole moments leading into charge separation. If the lattice of crystalline material is symmetric, these separated charges will cancel each other out and spontaneous polarisation cannot occur. In a lattice which lacks inversion symmetry a non-vanishing spontaneous polarisation is allowed. This is noteworthy especially when heterointerfaces are involved in the structure. The piezoelectric polarisation is formed when the crystal is strained. The strain unbalances the symmetry of the lattice and can lead into net charge separation and, thus, to electric fields inside the crystal. The strength and direction of the piezoelectric field is calculated from the strain and the piezoelectric tensor. Strain in the lattice usually occurs because of a lattice mismatch or by the difference in thermal expansion coefficients between the substrate and the epitaxial layer. Both spontaneous and piezoelectric polarisation (like external electric fields) contribute to bending of the band edges in a semiconductor structure leading into significant effects in device characteristics. For example, carrier dynamics and the functional layers in an LED structure may work in a different manner. Especially the QWs in an LED structure are affected by the polarisation fields. Figure 3.4 shows the effect of polarisation fields in an InGaN/GaN QW structure. Polarisation fields can twist a square shaped potential well into a triangular shape. Electrons and holes are therefore shifted on the opposite sides of the quantum well. This reduces the radiative recombination efficiency. Also the energy difference between the hole and the electron ground levels decreases leading into red shift of the emission wavelength. The effect is called quantum confined Stark effect (QCSE), which also occurs when the structure is biased with an external electric field. Polarisation can also affect the effective height of the barriers surrounding the well and, therefore, have an effect in leakage current. 3.4 Carrier lifetime in semiconductors Carrier lifetime is a quantity, which describes the mean time required for a charge carrier to recombine. Carriers can recombine radiatively, i.e., 14 Electrical properties of semiconductors Figure 3.4. Energy band diagrams of an InGaN/GaN quantum well a) without the effect of polarisation fields and b) with spontaneous (Psp ) and piezoelectric (Ppe ) polarisation fields. The band gap of InGaN QW reduces from E1 to effective band gap E2 . Image reprinted with permission from reference 30. releasing the freed energy as a photon. They can also recombine nonradiatively without releasing a photon. For example, the freed energy can be transferred to another charge carrier, a hole or an electron. If an electron captures the freed energy, it is pushed into a higher state in the conduction band. If a hole captures it, the hole is pushed into a deeper state in the valence band. The newly excited charge carrier gradually releases the obtained energy thermally and returns to the band edge. This type of recombination is called Auger recombination. Impurities and defects cause localized states in the band gap. These localized states can act as a middle step for electron transition from the conduction to the valence band. Since the states are localized, they are spread-out in the k-space. Therefore, the transitions through these localized states can emit phonons, instead of photons. A phonon is a quantum particle representing lattice vibrations, i.e., heat. This transition is typically non-radiative and also the dominant recombination mechanism in indirect band gap materials. It is also the dominant recombination mechanism for direct band gap materials, if the carrier concentration is low. This recombination is called Shockley-Read-Hall (SRH) recombination. These three recombination mechanism are the basis of the ABC model. The ABC model describes recombination rate, i.e., the inverse of lifetime. Recombination rate depends on the carrier concentration in a semiconductor and it is defined by three coefficients describing these three prominent recombination mechanisms. According to the ABC model, the recombination rate R as a function of carrier density n can be written as R(n) = An + Bn2 + Cn3 , (3.3) in which A is the coefficient for SRH, B for radiative and C for Auger recombination. 15 Electrical properties of semiconductors The lifetime of charge carriers defines the internal quantum efficiency (IQE) of the material or the device. IQE can be written according to the ABC model as Bn2 . (3.4) An + Bn2 + Cn3 Even though the ABC model is a simplified model and it lacks, e.g., the efIQE(n) = fect of leakage current, it has been a tool for understanding the operation and the efficiency of LEDs and LDs. The inverse of the recombination rate directly gives the lifetime of the charge carrier. Obtaining the coefficients for different recombination mechanisms has been one way to study the efficiency droop observed in the InGaN LEDs. Droop is discussed in more detail in chapters 4.2 and 5.4. According to the ABC model the recombination rates vary as functions of carrier density. Hence, in a LED the charge carrier lifetimes are strongly affected by the operating point, i.e., injection current and biasing of the LED. When measuring carrier lifetimes, usually the total recombination rate or lifetime is measured. Measuring a specific recombination lifetime requires measurement tricks for extracting the parameter. 3.5 Basic operating principle of a light emitting diode LEDs are typically fabricated from semiconductor materials, but also from organic materials. The LEDs fabricated from organic materials are called organic light emitting diodes (OLEDs). Even though the operating principles are similar in solid state LEDs and OLEDs, the OLEDs are not discussed in this work due to their differences in materials and in properties. The basic operating principle of a semiconductor LED can be explained through a simplistic interpretation of quantum mechanics described in previous sections. The structure is designed in a way that allows a controlled injection of both types of charge carriers, electrons and holes, into a region, which favours radiative recombination. This region is often referred to as the active region and it can be designed in various ways. The injection of holes and electrons is made possible by utilizing a pnjunction. In a conventional LED the active region is placed inside the pnjunction allowing electrons and holes to be injected into the active region from the n- and p-doped regions, respectively. The material used in the active region, i.e., the light emitting region, dictates the wavelength of the emitted light. As mentioned earlier, recombination releases a quantum of energy defined by the band gap of the material. If a photon is emitted, 16 Electrical properties of semiconductors its energy Eph defines its wavelength λ. The relation between the photon energy and wavelength is Eph = hc , λ (3.5) in which h is the Planck’s constant and c is the speed of light. Separate active region is in principle not necessary, but the efficiency in a pure pn-junction LED would be unreasonably low due to poor carrier confinement and low radiative recombination rate. The most basic practical active region scheme is a double heterojunction (DHJ) structure, in which a lower band gap material is sandwiched between the p- and nlayers fabricated from a higher band gap material. In III-nitride LEDs the active regions are quantum well (QW) based. A QW structure is a DHJ structure, in which the sandwiched layer is only a few nanometers thick. When considering QWs, the higher band gap layers are called barriers. QWs are often preferred over thicker films in active regions due to improved carrier confinement. Due to electric fields, whether they are external or internal, the holes and electrons tend to move to the opposite sides of the active region lowering the recombination rate. In a thin QW the distance of carrier separation can be lowered and, thus, improve the recombination efficiency. If the device has one QW it is typically called a single quantum well (SQW) LED. If the device has a periodic structure which is formed by repeating multiple QWs and barriers, it is called multi quantum well (MQW) LED. MQW structures are often preferred instead of SQW structures again due to improved confinement properties. In SQW structures, if charge carriers manage to escape the QW, they are lost for radiative recombination whereas in MQW structures the escaped carriers can be recaptured in another QW leading into less leakage current and higher light output powers. Additionally, having multiple QWs increases the volume of the active region and reduces the droop effect, which is discussed in more detail in sections 4.2 and 5.4. Figure 3.5 shows a schematic illustration of a GaN based SQW LED. As in a normal pn-junction, when the LED is biased, a small electric field in the p- and n-type regions transports majority carriers towards the depletion region. From the edge of the depletion region, the main current component is diffusion current. Consequently, typical LEDs are essentially 1D structures where diffusion transports electrons and holes to the active region from the p- and n-type regions located at opposite sides of the active region. However, charge carrier diffusion is one of the main causes for leakage current in LED structures. Diffusion can transport charge 17 Electrical properties of semiconductors Current spreading layer p-contact p-GaN n-contact DHJ QW n-GaN Figure 3.5. Schematic illustration of a GaN based SQW LED. The active region is sandwiched between the n- and p-type layers. The drift current, marked with arrows, is dominant transport mechanism for majority charge carriers in the p- and n-layers, whereas diffusion is the dominant transport mechanism in the depletion regions, i.e., near the active region and near the contacts. carriers relatively large distances and even over the active region [15,16]. Another source for leakage current is tunneling through the barriers of the QW structure. Diffusion induced electron leakage current in an LED can be estimated by the short diode law. The short diode law can be written as I = qApn qVa Dn np ∗ (e kB T − 1), L (3.6) in which q is the elementary charge, Apn the cross sectional area of the junction, Dn the minority diffusion constant of electrons, L the distance between the contact and the edge of the depletion region, np the minority carrier density at the edge of the depletion region without bias voltage, Va the applied bias, kB the Boltzmann coefficient, and T absolute temperaqVa ture. The last part, np ∗ (e kB T − 1), describes the excess minority carrier density at the edge of the depletion region. [31] The hole leakage can also be estimated in a similar manner, but in the case of a GaN pn-junction the hole leakage is negligible due to unbalanced hole and electron densities and hole transport being slower than electron transport. The efficiency of an LED can be defined in multiple ways depending on what effects are included into the definition. Three main efficiency parameters are typically used: internal quantum efficiency (IQE), external quantum efficiency (EQE) and wall plug efficiency (WPE). IQE is the most fundamental merit. IQE is defined as the ratio of generated photons per injected electrons and it describes mostly the quality of the active region or the device structure. EQE takes also into account the optical losses of the device, e.g., due to limited light extraction. It is defined as the ratio of photons extracted from the device per injected electrons. Ideally, EQE 18 Electrical properties of semiconductors equals to the product of extraction efficiency (EXE) and IQE. EXE describes the ratio of extracted photons per photons generated in the active region. WPE of an LED takes into account all the losses in the device, such as resistive losses in the material and in the contacts as well as the light extraction losses. It is defined as the ratio of emitted optical power per electrical power used to drive the device. Alternate definitions of IQE and EXE can be used in simulations, since these straightforward definitions of IQE and EXE do not take into account separately the effects of, e.g., photon recycling or injection efficiency ηinj . Injection efficiency describes the ratio of charge carriers recombining in the active region per injected charge carriers. In a well optimized conventional DHJ based LED ηinj is near unity. 19 Electrical properties of semiconductors 20 4. III-nitride light emitting diodes One of the key applications for III-nitride LEDs are white light sources. Since LEDs emit light with specific wavelength and with a reasonably narrow spectral width, yellow phosphors are used to down-convert part of the blue or UV emission from GaN based LEDs in order to obtain while light emission. A typical III-nitride LED has a n-doped GaN layer at the bottom, an active region in the middle and a p-doped GaN layer at the top. The order is dictated by the challenges in growth as discussed in section 5.2. Figure 4.1 shows a schematic illustration of a phosphor coated SQW InGaN LED. Figure 4.1. Schematic illustration of the structure of a white light LED. The structure has been grown on a sapphire substrate using an unintentionally doped buffer layer. The bottommost device layer is n-doped GaN following a single InGaN QW active region and a p-doped GaN layer. In order to improve current spreading a transparent metal thin film is deposited on the p-GaN layer. The LED chip is covered with down converting phosphor in order to convert blue emission from the QW into white light. 4.1 Properties of III-nitrides The device structures studied in this work were fabricated from III-nitrides. The group of III-nitrides is a subset of III-V semiconductors. The group 21 III-nitride light emitting diodes V atom is nitrogen as the name suggests. The III-nitride semiconductor family consists of aluminium, gallium and indium nitride (AlN, GaN, InN) and their alloys. There are also applications for boron nitride (BN) [32], but since it is not generally used in fabrication of LEDs, BN is not discussed in this work. AlN, GaN and InN are semiconductors with direct band gap making them suitable for light emitting applications such as LEDs and LDs [20]. GaN and GaN based semiconductors have been studied since the 1930’s. However, it was the breakthroughs in growth and successful p-type doping in mid-1990’s, which caused a large scale interest to study their possibilities and opened a door for a billion dollar business. GaN based semiconductor alloys are not only used in light emitting applications, but also in terahertz emitter applications, high power and high frequency electronics, solar power harvesting and light detection applications. The strong atomic bonding in GaN makes it mechanically and chemically stable as a material. It can also handle a large amount of radiation making it feasible for space applications. The band gap of AlN, GaN and InN is found in deep UV, in near UV and in infrared, respectively. Therefore, the band gap for their alloys can be, in theory, tuned from deep UV to infrared. [33] However, only the blue light emitters operate with high efficiency and if the alloy is tuned towards the longer or shorter wavelengths, the efficiency drops rapidly. In the deep UV range there are challenges with p-type doping [34] and large tensile stress due to the large lattice mismatch between the substrate and AlGaN films with high Al composition [22]. In AlGaN QWs used in UV LEDs dislocation density has much more impact in the quantum efficiency than in InGaN QWs used in blue light emitting LEDs [35]. In fact the blue light emitting InGaN alloy has a peculiar property to remain optically active despite large dislocation densities. The dislocation density in an InGaN film can be up to million times larger than in conventional IIIVs [36]. However, in InGaN QWs emitting at the longer wavelengths, the efficiency reduces again [37]. The origin of the low efficiency of the green InGaN light emitters is not perfectly understood [1]. However, defects acting as non-radiative recombination centres [38] and quantum confined Stack effect (QCSE) due to polarisation effects [39] are shown to be key elements in poor efficiency of high In content InGaN LEDs. High efficiency and high brightness light emitters are also challenging to fabricate from conventional III-Vs. The lack of high quality green LEDs and LDs is commonly called "the green gap". [40] 22 III-nitride light emitting diodes Even though sapphire is an electrically and thermally insulating material and a lattice mismatched substrate for III-nitrides, it has been the typical choice as the substrate material [20]. However, the need for a better substrate material has been a driving force for researchers and the industry to switch from sapphire to alternatives such as silicon carbide and bulk GaN [33]. Sapphire is still used for its cost efficiency and for the fact that it is suitable enough for fabricating devices with a reasonably high efficiency. In high power light emitting applications the improvements obtained by using, e.g., bulk GaN substrates overcomes the fabrication costs [41]. In addition to the mentioned materials III-nitrides can be grown on silicon substrates, which is beneficial in selected applications [42]. III-nitrides typically crystallise in the wurtzite crystal structure. The high ionic like behaviour in bonds between nitrogen and group III atoms and the polar crystal structure gives rise to piezoelectric properties for III-nitrides. [33] The piezoelectric properties can be exploited in fabrication of piezoelectric devices, but in the case of LEDs, the piezoelectricity brings challenges. The electric fields inside the crystal induce band bending causing problems in carrier transport and also affect the optical properties due to QCSE. [39] The piezoelectric fields in the crystal, whether they are spontaneous or strain induced, are always pointed in the c-direction, i.e., perpendicular to the basal plane. [33] This can be verified by simple arithmetic following Hooke’s law for the relation between stress and strain and from the geometry of the wurtzite crystal. III-nitrides are typically grown in c-direction. The challenges caused by the piezoelectric fields could be avoided, if the layer structure would be grown on so-called non-polar planes, e.g., m- and a-planes, which are oriented perpendicular to the c-plane. Also growth on tilted planes, i.e., semipolar planes, would reduce the effect of piezoelectric fields on the device performance. However, the crystal quality of films grown on non-polar and semi-polar planes is typically lower than in films grown on the c-plane. III-nitride growth can also be tuned to produce the zinc blende structure, which is a less polar structure. However, the zinc blende structure is a metastable crystal structure for III-nitrides and, thus, the wurtzite structure is preferred. 23 III-nitride light emitting diodes 4.2 Efficiency droop in III-nitride light emitting diodes Efficiency droop is a phenomenon, which has been gaining huge amount of interest in III-nitrides, in particular. All InGaN LEDs exhibit efficiency droop, i.e., reduction of quantum efficiency at high operating currents. Although droop has been under extensive study, several, even contradictory, theories have been proposed, such as density activated defect recombination (DADR) [43], Auger recombination [44], electron leakage [45], and unified models combining different mechanisms [3]. None of these have gained overall agreement in the research community [3]. A common feature in all the proposed models is the reduction of radiative recombination efficiency at high injection currents, or more specifically, at high carrier densities. Efficiency droop is not to be confused with the temperature related decrease of efficiency. The droop affects directly the IQE of the device and it is dependent on the carrier concentration in the active region. IQE defined by the ABC model can be fitted into EQE measurement data. However, it leaves significant uncertainties into the fitted parameters due to, e.g., lack of the effect of leakage current in the model. Moreover, the coefficient C describing Auger recombination rate obtained by fitting is usually unreasonably high, at least for the direct Auger process [46]. However, some studies indicate that high Auger recombination rates might be feasible [47]. It has also been suggested that the obtained coefficient C might not be describing only direct Auger recombination, but also defect or phonon assisted Auger recombination [46]. Much lower Auger coefficient, and also the droop effect, has been measured in nanowire LEDs compared to thin film LEDs [48]. Since nanowires can be grown practically defect free, this can be interpreted so that the Auger processes in III-nitrides are somehow related to the defects. Though Auger recombination most probably has a significant impact on the LED efficiency, other possible explanations, e.g., leakage current cannot be excluded. Typically, the reasons behind the droop mechanism can only be deducted from measurements by using mathematical models. Usually the evidence is not conclusive, because the models contain several parameters that can be fitted to the data. From an engineering point of view, pure Auger recombination as an explanation for droop would be the worst choice, because it is a phenomenon related to fundamental material properties and, therefore, difficult to reduce. If the reason would be defect 24 III-nitride light emitting diodes related Auger recombination, engineers could have some control over it by trying to reduce defect density. The best choice would be leakage current, because diminishing droop would become mostly an engineering problem. 25 III-nitride light emitting diodes 26 5. Experimental This chapter provides a short description of the sample fabrication by epitaxial growth for the research presented in this thesis. The standard lithography and etching methods are omitted due to their simplicity as well as the electrical and optical characterizations. In addition to epitaxy, this chapter presents the methods of measuring the droop effect as well as the carrier lifetimes and the recombination parameters in III-nitride LEDs. 5.1 Metal organic vapour phase epitaxy Metal organic vapour phase epitaxy (MOVPE), also known as metal organic chemical vapour deposition (MOCVD), is a gas phase deposition method for epitaxial growth of crystalline semiconductor materials. In epitaxial growth the film to be grown tries to copy the crystal structure and the orientation of the substrate. If the epitaxial film differs significantly from the substrate, it can grow as a polycrystalline film full of cracks, defects and stacking faults. However, with clever tricks it is possible to some extent grow material on a substrate, which has incompatible lattice structure, such as wurtzite GaN on silicon with the diamond crystal structure or on sapphire with the wurtzite crystal structure but a high lattice constant mismatch. The name for MOVPE stems from the source materials used in the growth process. In III-V growth the group III material precursors are metal organic compounds meaning molecules of metal atoms complexed with hydrocarbons. The typical precursors for group III in III-nitride growth are trimethylaluminium (TMA), trimethylgallium (TMG) and trimethylindium (TMI). The metal organics are stored in temperature controlled steel containers, called bubblers. The carrier gas, which is hydro- 27 Experimental gen, nitrogen or their mixture, is injected through the bubbler. The carrier gas absorbs molecules from the precursor and saturates. The saturated carrier gas is then diluted to obtain the preferred concentration of the precursor material. In MOVPE growth of III-nitrides, the source for group V atoms, i.e., nitrogen, is usually gaseous ammonia (NH3 ). The group III and group V precursors are directed into the growth chamber of the reactor through separate gas manifolds. They are mixed as late as possible to avoid pre-reactions of the precursors. The MOVPE system used in this work was a cold wall close coupled shower head (CCS) system, in which the precursors are fed from the top of the chamber through a shower head close to the heated susceptor. 5.2 Growth of III-nitride light emitting diode The structures grown for this work were grown on 2" c-plane sapphire wafers by using the two step nucleation technique [49–51]. In this method a thin film of GaN is grown at low temperature, at around 550 o C. The low temperature GaN is polycrystalline with a mixture of wurtzite and zinc blende GaN. After the desired thickness is obtained, the flows of the precursor gases are stopped and temperature is increased to around 1000 o C. The polycrystalline film starts to evaporate. Material with low crystal quality evaporates faster than material with high crystal quality. The precursor gases are introduced again into the reactor, when only few high crystal quality seeds are remaining on the substrate. The seed crystals start to grow larger and eventually they will coalesce. The coalescence points typically create threading dislocations, which will extend through the whole structure. When coalescence occurs, two-dimensional film growth begins. [51] Two-dimensional growth continues typically until a buffer layer with a thickness of a couple of micrometers is obtained, after which growth of the desired structure can be performed. The growth of the actual LED structure is typically performed on these unintentionally doped buffer templates [51]. The growth typically begins with the n-doped region. If a GaN film is grown on p-doped GaN, the Mg atoms tend to diffuse into the growing film making the growth of a sharp doping profile challenging [52]. This behaviour of Mg is called the memory effect. Mg atoms are known to reduce the optical quality of the InGaN alloy [53]. Therefore, if the InGaN active region would be grown directly on top of the p-GaN film, the Mg atoms diffusing into the active region 28 Experimental would reduce its optical efficiency. In addition, the surface morphology of p-doped GaN can be rough. Growing on a rough surface generates defects into the following films. However, the properties of the p-GaN layer can be affected by the carrier gas used during growth [54]. Growth of p-GaN in a hydrogen environment yields a significantly smoother surface than that in a nitrogen environment [55]. Moreover, challenges arise also in the post-growth processing, if the p-doped film is buried under the rest of the structure. The n-GaN layer is grown at 1000 o C. A structure called underneath layer is often grown on top of the n-doped layer. The underneath layer can be a film or a superlattice (SL) and its thickness varies. It typically contains a small amount of indium and it can be n-doped or unintentionally doped. In the structures grown for this research an unintentionally doped InGaN/InGaN SL underneath layer was used. The indium composition of the InGaN layers in the SL were 1 and 0.1 %, respectively. The underneath layer improves the performance of InGaN QWs by separating the active region from the growth interruption between high temperature n-GaN and reasonably low temperature InGaN [56]. It relieves strain and improves the morphology of the following layers [57, 58] as well as reduces threading dislocation density and improves the homogeneity of the QWs [59]. The growth temperature has to be lowered in order to obtain suitable growth conditions for the InGaN SL, since InN is less stable compared to GaN and too high growth temperatures lead into indium desorption. Thus, the underneath layer is grown at 750 o C. Following the underneath layer the InGaN QWs are grown. The growth temperature of the InGaN QWs is also 750 o C, but the barriers surrounding the wells are grown at slightly higher temperatures, at around 850 o C. Following the growth of the QW structure an electron blocking layer (EBL) is grown. EBL can be grown as a film or as an SL. It is typically fabricated from AlGaN alloy and it can be p-doped or unintentionally doped. The electron and hole transport properties differ significantly in III-nitrides and there is substantial difference in the hole and electron densities in the corresponding doped regions. Therefore, electrons tend to overflow the active region when the LED is operated. The EBL, as the name implies, is used to reduce the electron overflow. The AlGaN/GaN band alignment is such that it creates a potential barrier for electrons without significantly affecting hole transport. Growing good quality AlGaN requires high temperatures and thus, the growth temperature is 29 Experimental increased to 950 o C. For the structures used in this research, p-doped AlGaN/GaN SLs were used as EBLs. The aluminium content of the AlGaN layers was 12 %. Finally, the p-doped GaN layer is grown at 950 o C. A heavily doped surface is grown on top of the p-GaN layer in order to reduce the contact resistance of the p-contact, which is deposited during the processing of the LED chips. The heavily doped surface is achieved by flowing only Mg precursor in the reactor at 800 o C. Since the acceptor states of the Mg atoms are passivated by hydrogen impurities in the epitaxial film, the acceptors have to be activated [60]. This can be done by annealing, which can be done in-situ or ex-situ [27]. The annealing of the structures fabricated for this research was performed in-situ. Figure 5.1 shows schematically the final layer structure together with the metal contact layers. p-GaN EBL SQW Underneath n-GaN i-GaN GaN InGaN GaN Sapphire Figure 5.1. Schematic illustration of a GaN based SQW LED. The structure is grown on an i-GaN buffer layer on top of a sapphire wafer. The growth of the actual structure is began with n-doped GaN following the underneath layer, which is a InGaN/GaN SL with a very low indium content. The active region is an InGaN QW sandwiched between GaN barriers. Growth is continued with an AlGaN/GaN EBL SL, after which the p-GaN layer is grown. 5.3 Measuring carrier lifetimes Typically carrier lifetimes have been measured with the time resolved photoluminescence (TRPL) method [61–63]. In this method the studied structure is excited with extremely short (in the range of femtoseconds) laser pulses and the photoluminescence (PL) intensity is measured as a function of time. The carrier lifetime can be calculated from the decay curve of the PL intensity. The method can be considered to be reasonably accurate and it has been used for decades. With this method it is generally challenging to measure carrier lifetimes in a specific operating point, since the operating point is constantly changing during the decay of the PL intensity. In addition, the measurement becomes more challenging, when the operating point of interest would require low excitation conditions. 30 Experimental Another method to get an estimate for carrier lifetime is to utilize differential carrier lifetime analysis. This can be performed optically [64] or electrically [65]. These optical and electrical differential carrier lifetime analysis methods were used in Publication I, in which the carrier lifetimes were measured from near UV InGaN/GaN MQW LEDs as a function of injection current. The optical method is based on injecting the LED with a direct current (DC) component modulated by a sinusoidal alternating current (AC) component, which is small enough not to significantly affect carrier concentration or the operating point defined by the DC. The light output is then measured and it’s sinusoidal form is compared to the AC component of the injection. The carrier lifetime τ causes a phase difference φ between the AC injection and the optical output and their relation can be written as τ= tanφ , 2πf (5.1) in which f is the frequency of the AC component [64]. The phase difference can be obtained with various tools such as an oscilloscope or a lock-in amplifier. In the measurements performed for Publication I a lock-in amplifier was observed to be more accurate with a low DC injection. The electrical method relies on an equivalent circuit. Ideally, the electrical properties of a LED can be described in a specific operating point by lumped elements. The active region is described by defining a parallel connection of the differential resistance Rd and capacitance CLED . In addition, the resistive losses can be lumped into the series resistance Rs . The carrier lifetime τ can be obtained by τ = CLED Rd . (5.2) In addition to the equivalent circuit of the LED structure, the inductance of the test leads was taken into account by adding a series inductance Ls into the equivalent circuit. The equivalent circuit used in Publication I is shown in figure 5.2. First, the LED was characterized using an impedance analyser. The parameters CLED and Rd were then obtained by fitting the parameters of the equivalent circuit to the measurement data. Both measurement setups have some challenges. The electrical method is susceptible for resonance effects and the optical for phase shifts caused by the drive circuit. However, the measurements have reasonable agreement between each other and similar results obtained with TRPL can be found in the literature [61]. Significantly shorter lifetimes have also been 31 Experimental Figure 5.2. Equivalent circuit used in obtaining the carrier lifetime from the impedance analysis. Ls , Rs , Rd and CLED stand for inductance of the lead wires, series resistance of the contacts, differential resistance of the LED and capacitance of the LED, respectively. Figure modified from Publication I. reported in the literature by using TRPL [62, 63] and current pulsing [66] methods. However, comparing TRPL results with the results obtained in Publication I is challenging due to the nature of TRPL, in which the operating point of the studied structure is not well defined during the measurement. With low excitation the Shockley-Read-Hall (SRH) recombination should dominate the carrier lifetime [67]. Ryu, et al., estimated by studying the Auger recombination parameters found in literature, that the SRH lifetime should be in the range of 300 ns to 1500 ns [68]. This is in good correlation with the lifetimes obtained in Publication I. 5.4 Measuring droop The direct measurement of IQE would remove some ambiguousness compared to the measurement of EQE. However, measuring IQE is not straightforward and always requires some assumptions. One common method to measure IQE is to measure photoluminescence (PL) intensity at cryogenic temperatures and to compare it to the PL intensity measured at room temperature [69]. The principle in the method is based on the assumption that defect states in the device or film are passivated at the cryogenic temperatures. Then IQE at cryogenic temperature is 100 %. The ratio between PL intensities measured at room and cryogenic temperatures is directly the IQE of the device at room temperature. The method is reasonably straightforward, but it lacks the possibility to study the IQE as a function of injection current and requires measurements performed at cryogenic temperatures. With similar assumptions the measurements can be performed with the temperature dependent EL method [70]. IQE can also be determined without measuring the samples at cryogenic temperatures, e.g., by simulating the optics of the structure to remove the effect of optical losses from the measurement data [71, 72] or fitting the 32 Experimental rate equation parameters to the measurement data [73]. The optics simulation method is used to obtain the EXE of the device and using the obtained EXE to deduce the IQE from the EQE measurements. The method requires detailed knowledge on the optical behaviour of the structure in order to perform a valid simulation and, therefore, it can be used only in simple device structures. The rate equation fitting method requires several assumptions in order to achieve a good estimate for the charge carrier generation rate. In Publication II a simple method for determining IQE and EXE is studied. In contrast to conventional methods, the studied method does not require measurements to be performed at low temperature nor detailed knowledge of the device structure. In this method the optical output is measured with an integrating sphere using a calibrated spectrometer with a wide range of injection currents. From the measurement data EQE and its second derivative is calculated as a function of optical output power. The basic principle behind this is that the optical output power and the current density have differing dependency on the carrier density. Therefore, the calculated curves can be used to extract parameters for internal efficiency of a LED without detailed information about the structure itself. However, the method requires a good estimate for the Auger recombination coefficient. EXE can be obtained by measuring the maximum of the EQE and its second derivative in the vicinity of the maximum EQE. EXE can be then obtained by χ= 2 ηm , 2D ηm + 4Pm 2 (5.3) in which χ is EXE, ηm the maximum EQE, Pm the photo current at the maximum EQE conditions and D2 the second derivative of the EQE in the vicinity of the maximum EQE. The internal recombination parameters can then be obtained by A= and B= 1 − ηm /χPm qηm 2ηm C 1 − ηm /χ 2/3 2/3 3 C 4 Pm 1 · , qd χ (5.4) (5.5) in which q is the elementary charge and d the thickness of the active region. A, B and C are the recombination parameters for SRH, radiative and Auger recombination, respectively. 33 Experimental 34 6. Diffusion injected light emitting diode The following sections describe a fundamentally different approach to electrically excite the active region of light emitting devices based on bipolar diffusion of charge carriers. In contrast to the existing double heterojunction (DHJ) solutions, this approach allows to excite nanostructures located at device surfaces without the need of forming a DHJ, absorbing layers or even electrical contacts on the device surface. 6.1 Background The basic structure of LEDs and the associated current injection principle have remained essentially unchanged for decades. However, the conventional LED design faces challenges in charge carrier injection into near surface nanostructures or devices enhanced by near field plasmonics. Nanostructured DHJs are challenging to realize and typically lead into the formation of a depletion layer causing large leakage currents [5]. The development of optoelectronic structures, which have exceptional potential, such as photonic crystal [6], surface plasmon enhanced [5, 7–11], nanowire [4, 12, 13] and quantum dot [14] structures, is obstructed by these challenges. The theoretical background for diffusion injection was first developed by Kivisaari et al. The group showed through simulations, that the diffusion currents can be used to inject charge carriers to III-nitride nanowires. The results of the simulations were extremely promising with up to 95 % injection efficiency. The simulated nanowires were not truly nanowires, since their diameter were defined as 1 μm. Due to their large size, the quantum effects in carrier transport were neglected. [74] 35 Diffusion injected light emitting diode 6.2 Current Transport in DILED In a DILED structure the active region is located outside the pn-junction. The carrier injection into the active region is caused by bipolar diffusion of electron and holes. In contrast to conventional DHJ structures, both charge carriers enter the active region from the same side. When a DILED is biased, the electrons and holes are transported from the p- and n-type regions towards the depletion region. However, the charge carriers are affected by the low carrier concentration in the active region and diffuse towards the area with low carrier density as shown in the drift-diffusion equation 3.1. They are confined in the lower band gap material and recombine. The constant recombination in the active region maintains the low carrier concentration in the active region and, therefore, the active region acts as a drain for the charge carriers. Diffusion is a reasonably fast process, which allows the diffusion transport to occur without significant recombination in the pn-junction. The current transport in a DILED can be estimated with equation 3.6 by replacing A and L to reflect the DILED structure. Since both electrons and holes are transported from the same side and the electron and hole fluxes into the active region are equal, the net current into the active region is zero. However, due to the bipolar transport of charge carriers, the polarisation fields will always have a negative impact on the transport of either type of charge carriers into the active region. 6.3 Buried MQW DILED As a first experimental demonstration of diffusion injected light emitting diode (DILED), an InGaN/GaN MQW stack with a GaN pn-junction on top of it was grown, processed to a device and characterized optically and electrically. The structure for the first experimental demonstration was this buried MQW structure due to its simplicity in processing. Additionally, since the pn-junction is located on top of the MQW stack, there is no need to etch through the active region keeping it completely intact. The buried MQW DILED was studied in Publication III and in Publication IV. The samples were grown on top of 2" c-plane sapphire substrates with MOVPE. The structure consisted of a 3 μm thick unintentionally doped GaN (i-GaN), on top of which a five well InGaN/GaN MQW stack was grown, followed by a 100 nm n-doped GaN layer, a 30 nm i-GaN spacer 36 Diffusion injected light emitting diode layer, a p-doped AlGaN/GaN superlattice electron blocking structure, and a 400 nm p-type GaN layer. More details of the growth parameters can be found in chapter 5.2 and in Publications III and IV. Following the MOVPE growth the wafer was patterned using standard lithography methods. The device was designed to be a fingered comb-like structure. The n-type layer was revealed by using inductively coupled plasma-reactive ion etching (ICP-RIE). Ni/Au/Ti/Au contact was deposited as the p-contact and Ti/Au as the n-contact. Schematic illustration of the fabricated device is shown in figure 6.1. !""#$ %"#$ Figure 6.1. (a) Schematic illustration of the fabricated device with the layer structure and (b) the dimensions of the fingers. Figure reprinted from Publication IV. The sample was characterized with electrical and optical measurement setups at various temperatures. The current-voltage (I-V) characteristics were found to be comparable to a conventional III-nitride LED. Strong blue light emission corresponding to the band-to-band emission from the InGaN/GaN MQW stack was observed when the structure was biased indicating significant diffusion current into the MQW stack. A negligible amount of GaN band-to-band emission was also observed. A strong yellow band luminescence was also observed with low injection currents. The yellow band luminescence became negligible, when the injection current was increased. The spectrum measured from the device with 20 fingers having widths of 30 μm is shown in figure 6.2. The yellow band luminescence was most probably originating from shallow dopants in n-GaN [75–77], carbon impurities in i-GaN [78–80], and gallium vacancy complexes in i-GaN [77, 80]. Figure 6.3 shows yellow band luminescence intensity PY B as a function of square root of the quan- 37 Diffusion injected light emitting diode Normalized radiant flux (a) 1 ← 250 mA (x1) 0.8 0.6 ← 50 mA (x5) 0.4 ← 10 mA (x50) 0.2 0 400 500 600 Wavelength (nm) Normalized radiant flux (b) 1 o ← 100 C 0.8 o ← 60 C 0.6 0.4 ← 0 oC 0.2 0 700 400 500 600 Wavelength (nm) 700 Figure 6.2. Normalized radiant flux of a DILED (a) with different injection currents at 20 o C and (b) at different temperatures with an injection current of 10 mA. Figure reprinted from Publication IV. tum well emission intensity PQW measured at 40 o C and a correspond- ing linear fit. As can be seen the fit is nearly perfect. The measurement was performed at various temperatures and the behaviour is similar at all temperatures. The assumption that the emission power PQW is proportional to the square of the carrier density in the QW as the ABC model of recombination suggests, indicates that the yellow band recombination is directly proportional to carrier density. This suggests that the recombination mechanism behind the yellow band emission takes place through localized states giving rise to Shockley-Read-Hall (SRH) recombination like properties. The buried MQW DILED exhibited an unusual temperature dependency. With increasing temperature the emission intensity increased. This can be seen in figure 6.4. This is in stark contrast to conventional LEDs, in which the intensity typically decreases. In general, the probability of radiative recombination decreases with increasing temperature. Additionally the SHR recombination typically increases. These phenomena are also present in the studied structure. However, it was shown with simulations that in the DILED structure increasing temperature has a significant impact on the charge carrier injection. In Mg doped p-GaN the 38 Diffusion injected light emitting diode 250 mA 200 mA 150 mA 100 mA PYB (a.u) 40 30 20 50 mA 10 10 mA 0 5 10 15 20 25 30 PQW1/2 (a.u) Figure 6.3. Optical power of the yellow band emission as a function of square root of the QW emission power (asterisk) and a linear fit (solid line). Figure reprinted from Publication IV. activation energy of the acceptor state is reasonably high. Therefore, the portion of Mg atoms, which actually contributes to the p-type conductivity is reasonably small and increasing temperature increases that portion significantly. In the studied structure, the holes must travel from p-GaN through the n-GaN layer to reach the MQW stack. Increasing active acceptor density by increasing temperature increases the hole diffusion current through n-GaN. This effect is more significant than the negative effect of temperature on radiative recombination. It was also observed in the simulations that lowering the doping level of the n-GaN should increase the injection efficiency of the device. In Publication V a similar structure was studied with the exception of the active region. The MQW stack was changed from a five well stack into a three well stack each well emitting at different wavelengths. The topmost well, i.e., the closest to the pn-junction, had the lowest indium composition emitting at 390 nm and the bottommost had the highest indium composition emitting at 480 nm. The middle QW had the emission peak at 430 nm. Bipolar diffusion is strong enough to excite all three QWs. The topmost QW was found to be too shallow to confine charge carriers efficiently leading into low emission intensity. The middle well was found to emit the most light. The quality of the bottommost QW was not as good as that of the middle QW due to reasonably high indium composition. The spectrum of the multicolour buried MQW DILED is shown in figure 6.5. The emission of the multicolour DILED exhibited peculiar position dependency. The middle QW emission was strongest in the p-contact finger tips closest to the n-contact pad and the bottommost QW emission in the n-contact fingers closest to the p-contact pad. Figure 6.6 shows a micro- 39 Diffusion injected light emitting diode Measured output power (mW) (a) 3.75 100 oC 3.00 80 oC 60 oC 2.25 40 oC 20 oC 1.50 o 0 C 0.75 0.00 0 50 100 150 Current (mA) 200 250 1 2 Output power density (W/cm ) (b) 100 oC 0.8 80 oC 60 oC 0.6 40 oC o 20 C 0.4 0 oC 0.2 0 0 10 20 30 40 2 Current density (A/cm ) Figure 6.4. (a) Measured QW emission power of the MQW DILED as a function of injection current at different temperatures. (b) The simulated emission power density as a function of corresponding current density shows similar current and temperature behaviour as the measurement. Figure reprinted from Publication IV. 12000 Middle QW Intensity (arb.) 10000 8000 6000 4000 Bottom QW 2000 GaN 0 350 Top QW 400 450 Wavelength (nm) 500 Figure 6.5. Emission spectrum under 100 mA pulsed current injection at 20 o C. The four noticeable emission peaks can be seen: GaN band-to-band transition and the three QW peaks. Figure reprinted from Publication V. scope image, in which the position dependency is observed. With increasing injection current the emission of all QW emission peaks 40 Diffusion injected light emitting diode Figure 6.6. Microscope image from the multicolour buried MQW DILED. Figure reprinted from Publication V. increased. However, the topmost QW increased the least and eventually the peak disappeared in the envelope of the middle QW peak. At high injection currents the bottommost QW peak increased more rapidly and eventually it became higher than the middle QW peak. Figure 6.7 shows the emission spectrum of the multicolour MQW DILED with various DC injection currents. 2 440 mA 10 100 mA 50 mA 1 Intensity (a.u.) 10 20 mA 0 10 5 mA −1 10 −2 10 350 400 450 500 Wavelength (nm) 550 600 Figure 6.7. Emission spectrum under electrical injection using various currents ranging from 5 to 440 mA. Intensity is in logarithmic scale. Figure reprinted from Publication V. The rapid increase of the bottommost QW peak intensity is affected by the Joule heating, which occurs at high injection currents. By increasing temperature diffusion into the bottommost QW is increased in expense of the two other QWs. Figure 6.8 shows the effect of temperature to the emission peak intensities. 41 Diffusion injected light emitting diode Intensity (a.u.) (a) 1000 500 0 0 20 40 60 80 100 80 100 80 100 Temperature ( oC) (b) Intensity (a.u.) 10000 8000 6000 4000 0 20 40 60 Temperature ( oC) (c) Intensity (a.u.) 4000 3000 2000 1000 0 0 20 40 60 Temperature ( oC) Figure 6.8. Emission peak intensity of (a) the topmost QW, (b) the middle QW and (c) the bottommost QW. Increasing temperature decreases the intensity in the two topmost QWs, but increases the emission from the bottommost QW. Figure reprinted from Publication V. 42 Diffusion injected light emitting diode 6.4 Surface QW DILED The second generation DILED was studied in Publication VI. The DILED was fabricated with a near surface QW active region. The layer structure was grown on a c-plane 2" sapphire wafer with MOVPE. The growth process began with depositing the unintentionally doped GaN buffer layer, followed by Mg doped p-type GaN and Si-doped n-type GaN, which formed the pn-junction of the device. On top of the pn-junction an (In)GaN underneath superlattice was grown in order to improve the crystal quality of the following InGaN QW. The InGaN QW was capped with a 10 nm thick unintentionally doped GaN layer. More details of the growth parameters can be found in chapter 5.2 and in Publication VI. The device structure was patterned with standard lithography methods using ICP-RIE for etching the openings for n- and p-type layers. Finally, metal contacts were evaporated using the lift-off technique. The emitter mesa was a 30 μm × 30 μm square and the n- and p-layers were contacted with Ti/Al and Ni/ITO layers, respectively, also leaving a 2 μm gap to the edges of the n-GaN mesa and the emitter mesa. For easy handling of the device, the contact pads were large squares reaching towards the emitter mesa through narrow extensions. Figure 6.9 shows a schematic illustration of the structure of the surface QW DILED. Figure 6.9. Schematic illustration of the structure of the surface QW DILED. The emission spectrum of the device was measured using a μ-EL setup. Due to the small size of the emitter mesa, the overall light output of the device was reasonably low. Therefore, the optical power was measured indirectly by comparing the surface brightness to the surface brightness of a reference LED. The reference LED was previously characterised by using an integrating sphere. The reference LED was fabricated with similar doping levels and the similar active region as the surface DILED. The reference LED was patterned with standard lithography methods to 43 Diffusion injected light emitting diode chips with a surface area of 500 μm × 500 μm. A transparent Ni/Au current spreading layer was deposited on p-GaN and Ti/Au was used as the contact pad metallisations of the LED. The fabricated device was diced and attached to a heat sink. Current to the device contacts was injected through probe needles. The bias voltages in the near surface DILEDs generally exceeded 10 volts because the etching damage on the p-GaN surface made deposition of good quality contacts challenging. Moreover, the buried p-doped GaN layer was most likely not properly activated or it was re-passivated during the growth of the rest of the layer structure. Therefore, not much attention was paid to the I-V characteristics, which were dominated by the voltage drop at the contacts. However, according to the previous studies on diffusion injection, it is expected that the I-V characteristics of the device itself do not differ significantly from the conventional DHJ structures. The high bias voltages also resulted in significant Joule heating leading to relatively high local operating temperatures. Compared to the buried MQW DILED the current injection into the active region was significantly improved, which can be seen in higher surface brightness of the device as well as lack of yellow band emission. Figure 6.10 shows the measured spectrum as well as the optical output power and the normalized EQE. 20 mA Optical power (mW) Intensity (a.u.) 300 16 mA 200 12 mA 100 8 mA 0.6 0.4 0.2 0 0 5 10 15 Current (mA) 20 0 300 400 500 600 700 Wavelength (nm) 800 900 Figure 6.10. Emission spectrum, the optical output power and the external quantum efficiency (EQE) measured from the front side of the device. EQE was normalized by the maximum EQE of the reference LED. Figure modified from Publication VI. The emission from the surface QW had a peculiar position dependency along the plane of the mesa. This can be seen in figure 6.11. With small injection currents the emission is visible only at the edge of the QW mesa near the n-contact. Increasing the injection current the emission becomes more uniform and eventually the whole mesa is illuminating. Even though the doping levels, layer thicknesses or device geometry of 44 Diffusion injected light emitting diode Figure 6.11. Microscope images of the 30×30 μm2 surface QW under electrical excitation with injection currents of (a) 1 mA, (b) 5 mA, (c) 10 mA and (d) 20 mA. The emission intensity profile along (e) x and (f) y directions defined in a. With low injection the QW emits mostly near the n-contact. By increasing the injection current the emission becomes more uniform. Figure reprinted from Publication VI. the surface QW DILED were not optimized, it operated with one fifth of the EQE of the reference LED. The surface QW DILED is a conclusive evidence that the diffusion injection scheme can be utilized for excitation of surface or near surface light emitting structures. 6.5 Future prospects of DILED structures DILED is a completely novel type of LEDs that has potential to affect how LEDs are designed and fabricated. Diffusion injection is not limited to III-nitrides and possible applications can be found in conventional III-V semiconductors and III-Vs combined with silicon pn-junctions. Utilisation of diffusion currents is not limited to light emitting applications, but can also be used in solar cell devices and detectors. The principle has now been demonstrated to operate, even without proper optimization. In applications such as nanowire LEDs and surface plasmon enhanced LEDs, the diffusion injection has even more potential. Nanowires are considered as a promising way to improve the efficiency of a LED. Their superior crystal quality combined with low strain compared to thin films makes them the ultimate choice as an active region for LEDs [48]. Additionally, InGaN nanowires with a high indium composition reaching into the green and even to the red spectral range have been 45 Diffusion injected light emitting diode demonstrated [29]. Nanowires can be grown directly on top of the pnjunction giving rise to green solid-state lighting applications. Figure 6.12 shows a schematic illustration of a nanowire DILED structure. Figure 6.12. Schematic illustration of nanowire DILED structure. Diffusion injection removes the need for the top contact of the nanowire array. In the case of III-nitrides the n- or p-layer has to be etched for depositing the metal contact for the anode or cathode due to lack of proper dopant implantation techniques. In the case of silicon combined with conventional III-Vs, the pn-junction can be direcly implanted into the silicon substrate and the nanowires can be grown on the unetched plane of the pn-junction. Utilizing silicon pn-junction for excitation of surface active region has distinct possibilities in integrating light sources into conventional silicon technology perhaps even allowing all optical logic circuits fabricated on silicon wafers. Another interesting application for diffusion injection is to enable current injection into optically active 2D materials such as MoS2 [81]. Injecting electrons and holes by conventional sandwich-structure is cumbersome due to challenges in growth. Since diffusion injection allows the carriers to be transported from the same side, these 2D materials can be then deposited or transferred directly on top of the pn-junction. Due to relatively large refractive indexes of semiconductors, a large part of the emitted light is trapped in the device layer structure. With a properly designed metal grating, the wavelength of the surface plasmons in the metal can be tuned to be in resonance with photons generated in a nearby QW. This allows direct energy transfer from the QW to the surface plasmons of the metal grating. The surface plasmons are then able to radiate their energy from the surface of the device as photons. This direct coupling removes the limitation of total internal reflection at the semicon- 46 Diffusion injected light emitting diode ductor/air interface. However, in order to have coupling, the QW must be only a few nanometers away from the grating. The conventional DHJ approach is challenging, since the p- or n-layer on top of the QW would not be thick enough for proper excitation of the QW. By using diffusion injection, the QW can be placed near the surface naturally. Figure 6.13 shows a schematic illustration of a surface plasmon enhanced DILED structure. Figure 6.13. Schematic illustration of a surface plasmon enhanced DILED structure. By using diffusion injection, the active region, e.g., a SQW, can be placed close to surface for coupling into the surface plasmons of the nanopatterned metal grating on top of the device. In principle, diffusion injection should be more scalable than the conventional DHJ approach. This would give rise to the possibility of fabricating truly large area LEDs. The active region can be placed under the pn-junction and, therefore, it is not needed to etch through the active region. Fabricating large area LEDs could be one solution to overcome the droop effect, since the carrier density in the active region can be lowered significantly. 47 Diffusion injected light emitting diode 48 7. Conclusion Experimental studies on diffusion injection and its possibilities in light emitting applications have been discussed in this thesis. First, the history of LEDs and III-nitrides was reviewed. In addition, the studies on carrier lifetime measurements and the droop in III-nitride LEDs have been discussed due to their importance in understanding the behaviour of LEDs and DILEDs and the physics and limitations behind these devices. The fabricated DILED structures prove the feasibility of utilizing diffusion injection in excitation of light emitting materials. The electrons and holes can be injected from the same side of the active region, which removes the need of placing the active region in the conventional current path, i.e., inside the pn-junction. The main applications for diffusion injection are found in nanowire or other surface nanostructured devices, which are challenging to realize with the conventional DHJ approach. The efficiency of the first fabricated DILED device was in the range of the first blue LEDs fabricated from III-nitrides, even without any optimization in film thicknesses, the geometry or doping levels. Therefore, the buried MQW DILED structure has plenty of room for optimization and the demonstrated buried MQW DILEDs can be seen as a starting point for developing light emitting structures with large and continuous active region. The areally large active region reduces carrier density in the active region, which could be used to diminish the droop effect in IIInitride LEDs. The quantum efficiency of the second generation device, i.e., the near surface QW DILED, was already in the same order of magnitude as that of the reference LED, fabricated using the conventional DHJ approach. Both the first and second generation DILEDs have a lot of room for optimizing the structure, since neither layer thicknesses, doping levels nor the device geometry were optimized. According to simulations, a prop- 49 Conclusion erly designed nanowire DILED structure can obtain injection efficiency of near unity, which would give rise for DILED to challenge the conventional DHJ structure. Diffusion injection enables novel device structures, which cannot be realised by using the conventional current injection method. Such applications are nanowire LEDs including nanowire integration directly on silicon substrates, LEDs utilizing surface plasmon coupling and large area LEDs. 50 References [1] C. J. Humphreys, “Solid-state lighting,” MRS Bulletin, vol. 33, p. 459, 2008. [2] L. Jun-Lin, Z. Jian-Li, W. Guang-Xu, M. Chun-Lan, X. Long-Quan, D. Jie, Q. Zhi-Jue, W. Xiao-Lan, P. Shuan, Z. Chang-Da et al., “Status of GaN-based green light-emitting diodes,” Chin. Phys. B, vol. 24, p. 067804, 2015. [3] J. Piprek, “Efficiency droop in nitride-based light-emitting diodes,” Phys. Status Solidi A, vol. 207, p. 2217, 2010. [4] J. Wallentin, N. Anttu, D. Asoli, M. Huffman, I. Åberg, M. H. Magnusson, G. Siefer, P. Fuss-Kailuweit, F. Dimroth, B. Witzigmann, H. Q. Xu, L. Samuelson, K. Deppert, and M. T. Borgström, “InP nanowire array solar cells achieving 13.8% efficiency by exceeding the ray optics limit,” Science, vol. 339, p. 1057, 2013. [5] M.-K. Kwon, J.-Y. Kim, B.-H. Kim, I.-K. Park, C.-Y. Cho, C. C. Byeon, and S.J. Park, “Surface-plasmon-enhanced light-emitting diodes,” Advanced Materials, vol. 20, p. 1253, 2008. [6] J. J. Wierer, A. David, and M. M. Megens, “III-nitride photonic-crystal lightemitting diodes with high extraction efficiency,” Nature Photonics, vol. 3, p. 163, 2009. [7] V. J. Sorger and X. Zhang, “Spotlight on plasmon lasers,” Science, vol. 333, p. 709, 2011. [8] R. F. Oulton, V. J. Sorger, T. Zentgraf, R.-M. Ma, C. Gladden, L. Dai, G. Bartal, and X. Zhang, “Plasmon lasers at deep subwavelength scale,” Nature, vol. 461, p. 629, 2009. [9] D.-M. Yeh, C.-F. Huang, C.-Y. Chen, Y.-C. Lu, and C. C. Yang, “Localized surface plasmon-induced emission enhancement of a green light-emitting diode,” Nanotechnology, vol. 19, p. 345201, 2008. [10] K. Okamoto, I. Niki, A. Shvartser, Y. Narukawa, T. Mukai, and A. Scherer, “Surface-plasmon-enhanced light emitters based on ingan quantum wells,” Nature Mater., vol. 3, p. 601, 2004. [11] W. L. Barnes, A. Dereux, and T. W. Ebbesen, “Surface plasmon subwavelength optics,” Nature, vol. 424, p. 824, 2003. [12] D. Bayerl and E. Kioupakis, “Visible-wavelength polarized-light emission with small-diameter InN nanowires,” Nano Letters, vol. 14, p. 3709, 2014. 51 References [13] L. Cao, J. S. White, J.-S. Park, J. A. Schuller, B. M. Clemens, and M. L. Brongersma, “Engineering light absorption in semiconductor nanowire devices,” Nature Materials, vol. 8, p. 643, 2009. [14] M. G. B. Y. Shirasaki, G. J. Supran and V. Bulović, “Emergence of colloidal quantum-dot light-emitting technologies,” Nature Photonics, vol. 7, p. 13, 2013. [15] K. J. Vampola, M. Iza, S. Keller, S. P. DenBaars, and S. Nakamura, “Measurement of electron overflow in 450 nm InGaN light-emitting diode structures,” Appl. Phys. Lett., vol. 94, p. 061116, 2009. [16] H.-S. Kim, E. Brueckner, J. Song, Y. Li, S. Kim, C. Lu, J. Sulkin, K. Choquette, Y. Huang, R. G. Nuzzo, and J. A. Rogers, “Unusual strategies for using indium gallium nitride grown on silicon (111) for solid-state lighting,” Proc. Natl. Acad. Sci. U.S.A, vol. 108, p. 10072, 2011. [17] H. J. Round, “A note on carborundum,” Electrical world, vol. 49, p. 309, 1907. [18] O. V. Lossev, “Cii. luminous carborundum detector and detection effect and oscillations with crystals,” The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science, vol. 6, p. 1024, 1928. [19] R. D. Dupuis and M. R. Krames, “History, development, and applications of high-brightness visible light-emitting diodes,” Journal of Lightwave Technology, vol. 26, p. 1154, 2008. [20] E. F. Schubert, Light-Emitting-Diodes. Cambridge University Press, 2006. [21] S. C. Jain, M. Willander, J. Narayan, and R. V. Overstraeten, “III-nitrides: Growth, characterization and properties,” Journal of Applied Physics, vol. 87, p. 965, 2000. [22] I. Akasaki and H. Amano, “Breakthroughs in improving crystal quality of GaN and invention of the p-n junction blue-light-emitting diode,” Jpn. J. Appl. Phys., vol. 45, p. 9001, 2006. [23] I. Akasaki, “Key inventions in the history of nitride-based blue LED and LD,” Journal of Crystal Growth, vol. 300, p. 2, 2007. [24] H. Amano, M. Kito, K. Hiramatsu, and I. Akasaki, “P-type conduction in Mg-doped GaN treated with low-energy electron beam irradiation (LEEBI),” Japanese Journal of Applied Physics, vol. 28, p. L2112, 1989. [25] S. Nakamura, Y. Harada, and M. Seno, “Novel metalorganic chemical vapor deposition system for gan growth,” Applied Physics Letters, vol. 58, no. 18, p. 2021, 1991. [26] S. Nakamura and G. Fasol, The Blue Laser Diode. Springer, 1997. [27] S. Nakamura, T. Mukai, M. Senoh, and N. Iwasa, “Thermal annealing effects on p-type Mg-doped GaN films,” Japanese Journal of Applied Physics, vol. 31, p. L139, 1992. [28] Y. Narukawa, M. Ichikawa, D. Sanga, M. Sano, and T. Mukai, “White light emitting diodes with super-high luminous efficacy,” Journal of Physics D: Applied Physics, vol. 43, no. 35, p. 354002, 2010. 52 References [29] W. Guo, M. Zhang, A. Banerjee, and P. Bhattacharya, “Catalyst-free InGaN/GaN nanowire light emitting diodes grown on (001) silicon by molecular beam epitaxy,” Nano Letters, vol. 10, p. 3355, 2010. [30] S. Suihkonen, “Fabrication of InGaN quantum wells for LED applications,” Ph.D. dissertation, Helsinki University of Technology, 2008. [31] J. Singh, Semiconductor Optoelectronics: Physics and Technology. pore: McGraw-Hill, Inc., 1995. Singa- [32] C. Dean, A. Young, I. Meric, C. Lee, L. Wang, S. Sorgenfrei, K. Watanabe, T. Taniguchi, P. Kim, K. Shepard et al., “Boron nitride substrates for highquality graphene electronics,” Nature nanotechnology, vol. 5, p. 722, 2010. [33] H. Morkoç, Handbook of Nitride Semiconductors and Devices, Vol. 1: Materials Properties, Physics and Growth. Wiley-VCH Verlag GmbH & Co. KGaA, 2008. [34] J. Piprek, “Ultra-violet light-emitting diodes with quasi acceptor-free AlGaN polarization doping,” Optical and Quantum Electronics, vol. 44, p. 67, 2012. [35] M. Iwaya, S. Terao, T. Sano, S. Takanami, T. Ukai, R. Nakamura, S. Kamiyama, H. Amano, and I. Akasaki, “High-efficiency GaN/Alx Ga1-x N multi-quantum-well light emitter grown on low-dislocation density Alx Ga1-x N,” physica status solidi (a), vol. 188, p. 117, 2001. [36] C. J. Humphreys, “Does In form In-rich clusters in InGaN quantum wells?” Philosophical Magazine, vol. 87, no. 13, p. 1971, 2007. [37] S. Nakamura, M. Senoh, N. Iwasa, and S. Nagahama, “High-brightness ingan blue, green and yellow light-emitting diodes with quantum well structures,” Japanese Journal of Applied Physics, vol. 34, p. L797, 1995. [38] F. C.-P. Massabuau, M. J. Davies, F. Oehler, S. K. Pamenter, E. J. Thrush, M. J. Kappers, A. Kovács, T. Williams, M. A. Hopkins, C. J. Humphreys, P. Dawson, R. E. Dunin-Borkowski, J. Etheridge, D. W. E. Allsopp, and R. A. Oliver, “The impact of trench defects in InGaN/GaN light emitting diodes and implications for the "green gap" problem,” Applied Physics Letters, vol. 105, no. 11, p. 112110, 2014. [39] Y. Zhao, S. H. Oh, F. Wu, Y. Kawaguchi, S. Tanaka, K. Fujito, J. S. Speck, S. P. DenBaars, and S. Nakamura, “Green semipolar (2 0 2̄ 1̄ ) InGaN lightemitting diodes with small wavelength shift and narrow spectral linewidth,” Applied Physics Express, vol. 6, no. 6, p. 062102, 2013. [40] J. J. Wierer and J. Y. Tsao, “Advantages of III-nitride laser diodes in solidstate lighting,” physica status solidi (a), vol. 212, no. 5, p. 980, 2015. [41] C. A. Hurni, A. David, M. J. Cich, R. I. Aldaz, B. Ellis, K. Huang, A. Tyagi, R. A. DeLille, M. D. Craven, F. M. Steranka, and M. R. Krames, “Bulk GaN flip-chip violet light-emitting diodes with optimized efficiency for highpower operation,” Applied Physics Letters, vol. 106, no. 3, p. 031101, 2015. [42] A. Dadgar, “Sixteen years GaN on Si,” physica status solidi (b), vol. 252, no. 5, p. 1063, 2015. 53 References [43] J. Hader, J. V. Moloney, and S. W. Koch, “Density-activated defect recombination as a possible explanation for the efficiency droop in GaN-based diodes,” Appl. Phys. Lett., vol. 96, p. 221106, 2010. [44] Y. C. Shen, G. O. Mueller, S. Watanabe, N. F. Gardner, A. Munkholm, and M. R. Kramer, “Auger recombination in InGaN measured by photoluminescence,” Appl. Phys. Lett., vol. 91, p. 141101, 2007. [45] 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, p. 183507, 2007. [46] J. Hader, J. V. Moloney, B. Pasenow, S. W. Koch, M. Sabathil, N. Linder, and S. Lutgen, “On the importance of radiative and Auger losses in GaN-based quantum wells,” Applied Physics Letters, vol. 92, p. 261103, 2008. [47] J. Iveland, L. Martinelli, J. Peretti, J. S. Speck, and C. Weisbuch, “Direct measurement of auger electrons emitted from a semiconductor lightemitting diode under electrical injection: Identification of the dominant mechanism for efficiency droop,” Phys. Rev. Lett., vol. 110, p. 177406, 2013. [48] W. Guo, M. Zhang, P. Bhattacharya, and H. Junseok, “Auger recombination in III-nitride nanowires and its effect on nanowire light-emitting diode characteristics,” Nano Letters, vol. 11, p. 1434, 2011. [49] I. Akasaki, H. Amano, Y. Koide, K. Hiramatsu, and N. Sawaki, “Effects of AlN buffer layer on crystallographic structure and on electrical and optical properties of GaN and Ga1-x Alx N (0 < x ≤0.4) films grown on sapphire substrate by MOVPE,” Journal of Crystal Growth, vol. 98, p. 209, 1989. [50] S. Nakamura, “GaN growth using GaN buffer layer,” Japanese Journal of Applied Physics, vol. 30, no. 10A, p. L1705, 1991. [51] T. Lang, “Fabrication of heteroepitaxial templates for GaN-based optoelectronic devices,” Ph.D. dissertation, Helsinki University of Technology, 2007. [52] H. Xing, D. S. Green, H. Yu, T. Mates, P. Kozodoy, S. Keller, S. P. DenBaars, and U. K. Mishra, “Memory effect and redistribution of Mg into sequentially regrown GaN layer by metalorganic chemical vapor deposition,” Japanese Journal of Applied Physics, vol. 42, no. 1R, p. 50, 2003. [53] S.-N. Lee, H. Paek, J. Son, H. Kim, K. Kim, K. Ha, O. Nam, and Y. Park, “Effects of Mg dopant on the degradation of InGaN multiple quantum wells in AlInGaN-based light emitting devices,” Journal of Electroceramics, vol. 23, p. 406, 2009. [54] O. Svensk, S. Suihkonen, T. Lang, H. Lipsanen, M. Sopanen, M. Odnoblyudov, and V. Bougrov, “Effect of growth conditions on electrical properties of mg-doped p-gan,” Journal of crystal growth, vol. 298, p. 811, 2007. [55] B. Schineller, A. Guttzeit, P. Lim, M. Schwambera, K. Heime, O. Schön, and M. Heuken, “The influence of Mg-concentration and carrier gas on the electrical and optical properties of gan:mg grown by MOVPE,” Journal of Crystal Growth, vol. 195, p. 274, 1998. 54 References [56] T. Akasaka, H. Gotoh, T. Saito, and T. Makimoto, “High luminescent efficiency of InGaN multiple quantum wells grown on InGaN underlying layers,” Applied Physics Letters, vol. 85, p. 3089, 2004. [57] N. Nanhui, W. Huaibing, L. Jianping, L. Naixin, X. Yanhui, H. Jun, D. Jun, and S. Guangdi, “Improved quality of InGaN/GaN multiple quantum wells by a strain relief layer,” Journal of Crystal Growth, vol. 286, p. 209, 2006. [58] ——, “Enhanced luminescence of InGaN/GaN multiple quantum wells by strain reduction,” Solid-State Electronics, vol. 51, p. 860, 2007. [59] P. T. Törmä, O. Svensk, M. Ali, S. Suihkonen, M. Sopanen, M. A. Odnoblyudov, and V. E. Bougrov, “Effect of InGaN underneath layer on MOVPEgrown InGaN/GaN blue LEDs,” Journal of Crystal Growth, vol. 310, p. 5162, 2008. [60] W. Götz, N. M. Johnson, J. Walker, D. P. Bour, H. Amano, and I. Akasaki, “Hydrogen passivation of Mg acceptors in GaN grown by metalorganic chemical vapor deposition,” Applied Physics Letters, vol. 67, p. 2666, 1995. [61] D. M. Graham, A. Soltani-Vala, P. Dawson, M. J. Godfrey, T. M. Smeeton, J. S. Barnard, M. J. Kappers, C. J. Humphreys, and E. J. Thrus, “Optical and microstructural studies of InGaN/GaN single-quantum-well structures,” J. Appl. Phys., vol. 97, p. 103508, 2005. [62] J. K. Son, S. N. Lee, T. Sakong, H. S. Paek, O. Nam, Y. Park, J. S. Hwang, J. Y. Kim, and Y. H. Cho, “Enhanced optical properties of InGaN MQWs with InGaN underlying layers,” Journal of Crystal Growth, vol. 287, p. 558, 2006. [63] S.-R. Jeon, S.-J. Lee, S. H. Jung, S. H. Lee, J. H. Baek, H. Jeong, O. H. Cha, E.-K. Suh, and M. S. Jeong, “Effect of V-shaped defects on structural and optical properties of AlGaN/InGaN multiple quantum wells,” J. Phys. D: Appl. Phys., vol. 41, p. 132006, 2008. [64] A. David and M. J. Grundmann, “Droop in InGaN light-emitting diodes: A differential carrier lifetime analysis,” Appl. Phys. Lett., vol. 96, p. 103504, 2010. [65] G. E. Shtengel, D. A. Ackerman, and P. A. Morton, “True carrier lifetime measurements of semiconductor lasers,” Electronic Letters, vol. 31, p. 1747, 1995. [66] A. Laubsch, W. Bergbauer, M. Sabathil, M. Strassburg, H. Lugauer, M. Peter, T. Meyer, G. Büderl, J. Wagner, N. Linder, K. Streubel, and B. Hahn, “Lumiscence properties of thick InGaN quantum-wells,” Phys. Status Solidi C, vol. 52, p. 5885, 2009. [67] W. G. Scheibenzuber, U. T. Schwarz, L. Sulmoni, J. Dorsaz, J.-F. Carlin, and N. Grandjean, “Recombination coefficients of GaN-based laser diodes,” J. Appl. Phys., vol. 109, p. 093106, 2011. [68] H.-Y. Ryu, H.-S. Kim, and J.-I. Shim, “Rate equation analysis of efficiency droop in InGaN light-emitting diodes,” Appl. Phys. Lett., vol. 95, p. 081114, 2009. 55 References [69] S. Watanabe, N. Yamada, M. Nagashima, Y. Ueki, C. Sasaki, Y. Yamada, T. Taguchi, K. Tadatomo, H. Okagawa, and H. Kudo, “Internal quantum efficiency of highly-efficient Inx Ga1-x N-based near-ultraviolet light-emitting diodes,” Appl. Phys. Lett., vol. 83, p. 4906, 2003. [70] G. Chen, M. Craven, A. Kim, A. Munkholm, S. Watanabe, M. Camras, W. Götz, and F. Steranka, “Performance of high-power III-nitride light emitting diodes,” Phys. Status Solidi A, vol. 205, p. 1086, 2008. [71] A. Getty, E. Matioli, M. Iza, C. Weisbuch, and J. S. Speck, “Electroluminescent measurement of the internal quantum efficiency of light emitting diodes,” Appl. Phys. Lett., vol. 94, p. 181102, 2009. [72] E. Matioli and C. Weisbuch, “Direct measurement of internal quantum efficiency in light emitting diodes under electrical injection,” J. Appl. Phys., vol. 109, p. 073114, 2011. [73] Q. Dai, M. F. Schubert, M. H. Kim, J. K. Kim, E. F. Schubert, D. D. Koleske, M. H. Crawford, S. R. Lee, A. J. Fischer, G. Thaler, and M. A. Banas, “Internal quantum efficiency and nonradiative recombination coefficient of GaInN/GaN multiple quantum wells with different dislocation densities,” Appl. Phys. Lett., vol. 94, p. 111109, 2009. [74] P. Kivisaari, J. Oksanen, and J. Tulkki, “Current injection to free-standing III-N nanowires by bipolar diffusion,” Appl. Phys. Lett., vol. 103, p. 031103, 2013. [75] I.-H. Lee, I.-H. Choi, C. R. Lee, and S. K. Noh, “Evolution of stress relaxation and yellow luminescence in GaN/sapphire by Si incorporation,” Appl. Phys. Lett., vol. 71, p. 1359, 1997. [76] S. Li, C. Mo, L. Wang, C. Xiong, X. Peng, F. Jiang, Z. Deng, and D. Gong, “The influence of Si-doping to the growth rate and yellow luminescence of GaN grown by MOCVD,” J. Lumin., vol. 93, p. 321, 2001. [77] M. A. Reshchikov and H. Morkoç, “Luminescence properties of defects in GaN,” J. Appl. Phys., vol. 97, p. 061301, 2005. [78] J. L. Lyons, A. Janotti, and C. G. V. de Walle, “Carbon impurities and the yellow luminescence in GaN,” Appl. Phys. Lett., vol. 97, p. 152108, 2010. [79] A. Kakanova-Georgieva, U. Forsberg, and E. Janzén, “Carbon-tuned cathodoluminescence of semi-insulating GaN,” Phys. Status Solidi A, vol. 208, p. 2182, 2011. [80] F. J. Xu, B. Shen, L. Lu, Z. L. Miao, J. Song, Z. J. Yang, G. Y. Zhang, X. P. Hao, B. Y. Wang, X. Q. Shen, and H. Okumura, “Different origins of the yellow luminescence in as-grown high-resistance GaN and unintentionaldoped gan films,” J. Appl. Phys., vol. 107, p. 023528, 2010. [81] K. F. Mak, C. Lee, J. Hone, J. Shan, and T. F. Heinz, “Atomically thin mos2 : A new direct-gap semiconductor,” Phys. Rev. Lett., vol. 105, p. 136805, 2010. 56 A al t o D D2 1 3/ 2 0 1 5 9HSTFMG*agfibg+ I S BN9 7 89 5 2 6 0 6 5 81 6( p ri nt e d ) 7 89 5 2 6 0 6 5 82 3( p d f ) I S BN9 I S S N L1 7 9 9 4 9 34 I S S N1 7 9 9 4 9 34( p ri nt e d ) I S S N1 7 9 9 4 9 4 2( p d f ) A a l t oU ni v e r s i t y S c h o o lo fE l e c t r i c a lE ng i ne e r i ng D e p a r t me nto fM i c r o -a ndN a no s c i e nc e s w w w . a a l t o . f i BU S I N E S S+ E C O N O M Y A R T+ D E S I G N+ A R C H I T E C T U R E S C I E N C E+ T E C H N O L O G Y C R O S S O V E R D O C T O R A L D I S S E R T A T I O N S
