GaN-based LEDs

Basically, the blue GaN-based LED confronts some severe problems lowering the efficiency and hindering the realization of solid state lighting. Lack of suitable substrate GaN epitaxy is the most important issue. Nowadays, GaN material was grown on sapphire susbtrate, which has a 15% smaller lattice constant than GaN, and different thermal expansion coefficient. That leads to a very high defect density and cracking of the layers when the structures are cooled down after growth. The problem was firstly solved by Amano and Akasaki by designing and growing a AlN buffer layer in 1986 [10]. Also, Nakamura grew GaAlN buffer layers on top of sapphire in 1991 [11] which make it possible to grow GaN on sapphire. In addition to the invention of buffer layer, Prof. S. Nakamura also solved the high growth temperature

problems by his two-flow growth reactor which opens the door of high quality GaN material on sapphire [12]. The third problem of for GaN-based LED is p-type doping.

Every semiconductor lighting device needs p-n junctions. Previous to Akasaki’s work p-type doping of GaN was impossibly. Akasaki (1988 at Nagoya University) found that samples after Low Energy Electron Beam Irradiation treatment (LEEBI) showed p-type conductivity [13-14]. Thus Akasaki demonstrated that in principle p-type doping of GaN compounds was possible. Nakamura was then found the solution to the puzzle of p-type doping. He found that previous investigators had annealed the samples in Ammonia (NH3) atmosphere at high temperatures.Ammonia dissociates above 400 oC, producing atomic hydrogen. Atomic hydrogen passivates acceptors, so that p-type characteristics are not observed. Nakamura solved this problem by annealing the samples in Nitrogen gas, instead of Ammonia [15]. Benefit by the effort of such pioneers, the blue GaN-based light emitting diode (LED) is now successfully commercialized. The typical structure of a blue GaN-based LED is illustrated in Figure 1.2. The layer of n -type GaN contains an excess of electrons,

whereas the p-type layer is a region from which electrons have been removed (i.e., in which “holes” have been formed). If a forward bias is applied, electrons and holes can

recombine, releasing energy in the transition layer in the form of light. The energy of the photon corresponds to the voltage bias in the transition region (the “bandgap”).

Sapphire and silicon carbide are often used as substrates, which allow for large-area heteroepitaxial growth.

1-3 Motivation

Typical InGaN/GaN LED are characterized by a substantial decrease in efficiency as injection current increases. This phenomenon which known as efficiency droop is a severe limitation for high power devices that operate at high current densities and must be overcome to enable the LEDs needed for solid-state general illumination. The efficiency droop is caused by a nonradiative carrier loss mechanism, which is small at low currents but becomes significant for high injection currents.

Competition between radiative recombination and this droop-causing mechanism results in the reduction in efficiency as current increases. The physical origin of efficiency droop remains controversial, and several different mechanisms have been suggested as explanations, including carrier leakage from the active region [16].

Auger recombination [17], junction heating [18], and carrier delocalization from In-rich low-defect-density regions at high carrier densities [19]. Carrier leakage in GaInN LEDs generally refers to the escape of electrons from the active region to the p-type region. These leakage electrons may then recombine with holes either in the p-type region or at the contacts, dominantly by nonradiative processes. Necessarily,

therefore, fewer holes than electrons are injected into the active region. These two phenomena that escape of electrons form the active region and reduced hole concentration of any carrier leakage explanation for droop. Hole injected into the active region may be the limiting factor, possibly due to the low p-type doping efficiency or the electron blocking layer (EBL) acting as a potential barrier also for holes. As a result of the low hole injection, current across the device is dominated by electrons. Devices with p-type active regions which should increase hole injection efficiency have been proposed as a solution to this problem.

However, it is not clear which cause the efficiency droops at high current. For this study, we investigated the excitation power dependence PL intensity at room temperature and temperature dependent intensity to confirm the confinement of carrier in different structure. Then we discussed the normalized efficiency as a function of injection current density at room temperature clearly and used APSYS simulation to make sure our model is correct, so the physical mechanisms of current dependent efficiency of InGaN/GaN LED has been confirmed.

This thesis is organized in the following way: In chapter 2, we give some theoretical backgrounds and characteristics about InGaN/GaN MQW structures. The experimental setups and theory are stated in chapter 3. In chapter 4, we present the experiment results and discuss for optical and electrical properties of InGaN/GaN

MQW LED with LT-GaN, InGaN/GaN and without prestrain layer. In chapter 5, we show the experiment results and discuss for physical mechanisms of graded quantum wells as a function of injection current density in InGaN/GaN LEDs. Finally, we gave a brief summary of the study in chapter 6.

Figure 1.1 The bandgap diagram of compound semiconductor materials.

Figure 1.2 The schematic of typical p-side up GaN-based LED