Motivation

Nearly, compared with conventional quantum well (QW) structure, the fabrication of quantum dot (QD) with three dimensional confinements of carriers and delta-like function attract a lot of attention. In this thesis, the InGaN QD was grown by antisurfactant method. We investigated the structure and optical properties for different size QD and developed the simulation to study the energy transition.

Moreover, we investigated the effect of thermal annealing for different size QD with a series annealing temperature. We also provide a theoretical simulation to analyze the mechanism of the InGaN/GaN QDs of thermal annealing.

Nonpolar structure is fabricated in order to increase the internal quantum efficiency.

It is not affected by polarization effect based on the direction of Ga and N atoms arrange normal to the growth axis. In this thesis, a-plane and c-plane MQWs with the same growth condition were prepared to be compared in their optical characteristic.

Nonpolar a-plane GaN base material grown on r-plane sapphire substrates which\

always accompany with a wavy, stripe-like growth feature possess a large density of threading dislocations and stacking faults. It is result from serious aniostropic in-plane strain between different crystal axis [53]. Recently, successful epitaxial lateral overgrowth (ELOG) of a-plane GaN on r-plane sapphire has been reported. ELOG not only improves significantly the material quality by reducing the density of threading dislocations but also alleviates the strain-related surface roughening and faceting [54-55]. Despite the ELOG assisted morphology and quality improvements in a-plane GaN over r-plane sapphire, the study of the epilayer quality and dislocations distribution in the ELOG epilayer is quite not lucid. In this letter, we

successfully improve [1120] a-plane GaN quality by using TELOG and the optical and structural properties is presented explicitly.

The organization of this thesis is as following: the analyses of nanostructure optical properties were investigated. In the chapter 2, we discuss the characteristic of InGaN based nanostructure, the simulation and growth methods of InGaN QD. The experiment principle and set up are described in the chapter 3. The detailed experimental results, theoretical analysis and the effects of thermal annealing for InGaN QD are presented in the chapter 4. In the chapter 5, we study the optical properties of a-plane InGaN/GaN MQW and compared with the behavior of c-plane samples. In the chapter 6, the TELOG was introduced to improve the sample quality.

Finally, we give a conclusion of the thesis in the chapter 6.

D(E)

E E

D(E) D(E) D(E)

2-D

3-D 1-D 0-D

Fig 1.1.1 Dimension of nanostructure and corresponding density of state

Fig 1.1.2 The carrier behavior in three dimensional confinement structure

Fig 1.2.1 Different plane and orientation of hexagonal GaN

Chapter 2 Theory of GaN-based nanostructure

There are three main methods to from the quantum dot (QD) structure including Stranski-Krastanow (S-K) growth, antisurfactant method, and selective method. We summarize the growth methods in the recent years and introduce below:

(i) Stranski-Krastanow (S-K) growth mode (layer-then island)

The evolution of an initially two-dimensional growth in to a three-dimensional growth front is a well-known phenomenon and has been frequently observed in various systems. This growth mode used by various materials grown under compressive stress on heterostructure by strain-driven QD formation during heteroepitaxy as a bottom-up approach. After deposition of a few two-dimension monolayers (MLs), island structures are self-formed on a 2D wetting layer as a result of the transition of the growth mode and the stress needs to be released, as shown in Figure.2.1.1. The strain relaxation mechanisms would be first published by Stranski and Krastanow so that we called the growth mode as S-K growth mode. Especially, S-K growth has been successfully demonstrated to form self-assembled quantum dot (SAQD) on large area substrate with a good throughput and compatibility to current microelectronic technology.

(ii) Antisurfactant method

Using this growth method, the surfactant is believed to play an important role in changing the surface free energy of the sample. We can use the unequal equation to show the surface free energy for three-dimensional growth mode in a system,

δs < δf + δi ,

where δs is the surface free energies of substrate, δf shows the surface free energies of film, and δi the surface free energies of film. As a third element like the surfactant is added, the sign of the unequal equal equation would be change by altering the

substrate surface free energy [16]. Using this growth method, the surfactant is believed to play an important role in changing the surface free energy of the material.

Some reports have shown the similar reversed effect occurs in GaN-based system [17-18]. The antisurfactant is believed to inhibit the film growth and intentionally modify the two-dimensional mode into-three dimensional mode. However, the role of the antisurfactant is still unknown in the element of island growth, therefore, carrying out the basic mechanism in growth way would be important for us in further detailed studies. In this thesis, antisurfactant method was used to grow InGaN/GaN QDs for the studies. The time of SiN treatment was used to control the size of quantum dot.

The detail will be discussed in later chapter.

(iii) Selective method

Self-assembling growth is a convenient method to get QDs but the position of QDs is difficult to be controlled. Using selective method to obtain QDs, in the contrast, the shape, size, and the position of the QDs could be artificially designed and controlled.

In general, the selective method can control the position of the QDs by the methods of focused ion-beam (FIB) irradiation and photo-assisted wet chemical etching. S. Sakai, et al [19] grew the QDs on Si-patterned GaN/sapphire substrates.

We could use different patterns to grow QDs in different shapes. Here, we simply introduce the InGaN QDs formation by selective method published by Y. Arakawa, et al [20]. In this thesis, after depositing three periods of InGaN MQWs on the grid-like

SiO2-GaN-saphirre patterned substrates, they believe that InGaN QDs structures are formed at the tops of the hexagonal pyramids.

Fig.2.1.2 shows a SEM top view and cross-section view of InGaN QDs structures.

Shown as Fig 2.1.2 (b), no material as deposited on the SiO2 mask and the position of the QDs were controlled very well.

2.2 Quantum confinement effect in semiconductor nanostructure

Quantum mechanism provide us an understanding of atoms, molecular, atomic nuclei and aggregation of them have both wave and partial properties within them.

So-called quantum confinement effect is considering a particle confined in a finite potential whose size can be compared with its’ wavelength. Let us consider the exciton in the well. If the well width is much small than the Bohr radius in bulk, no obvious confine energy is expected, because the exciton feel the same environment as in bulk. (The binding energy, Bohr radius and relative factors of bulk III-V semiconductor are listed in table 2.2.1.) If we decrease the well width at the same order or less than the exciton Bohr radius, the wavefuction and electron and hole confine in the potential well. The spatial wavefuction will be compressed in the well and it leads to enhancement of the exciton binding energy. This behavior has been demonstrated both in experiment and in theory for III-V semiconductors [21]. A brief equation given for the Hamiltonian describing the relative motion of exciton is

( , , ) ( , , ) ( ) where and are the Hamiltonian describing the electron and hole motion confined in the well. And

He H_h

e h( ) H ₋ rv

is Hamiltonian describing the Coulomb potential between electron and hole. The x-y-z coordinate is defined by the spatial confinement, and denoted the relative position vector between electro and hole. rv

Then the eigenvalue can be represent given by

e h

n n n e h

E =E +E −E ₋ Eq. (2.1.2) where E_n^e and E_n^h is the energy of nth quantum confined state, and is the binding energy of exction.

Ee h₋

The quantum confinement increase with reduction of the well width. In QW structure, one dimensional confinement is considered and the motion of exciton is free in the other two dimensions. For QD structure, with reducing the degree of freedom,

exciton is confined in three dimensional potential well. Thus the density of state is present in delta function. The electronic states are quantized and the energy levels become discrete. Because of the localization of carriers trapped at QDs structure where the energy separated between ground state and the first higher exciton state is larger than the thermal energy T, the lifetime of the 0-D exciton strongly is almost independent of temperature.

Kb

2.3 Simulation of quantum confined Stark effect by using the FEMLAB

FEMLAB is attractive software for modeling and simulating scientific and engineering problems based on using finite element analysis to solve partial differential equations (PDE). In this thsis, the simulation was taken into account three dimensional quantum confinements for QD structure. The energy states in QDs were simply solved by self-consistent Schodinger equation. We use cylindrical symmetric coordinate to describe the QD structure for simplifying 3D problem into 2D. The general form of the equation is described as fallow,

&& where is the effective mass of the well and barrier depending on different region we considered, is the eigenfuction of electron or hole.

e( ) m r

( )r Ψ

For the symmetry, the function can be described by cylindrical from

2 2 and the wavefunction can be separated by

( , , )r z ϕ χ( , ) ( )z r

Ψ = Θϕ Eq. (2.3.3) We can get a general solution is

exp[ilϕ]

Θ = Eq. (2.3.4)

where denote the quantum number of different energy levels. And we substitute the solution in the original equation and simplify it as fallow,

l

Finally, we can use this equation and substitute the corresponding factor in the general form of software to obtain the eigenfuction and eigenvalue. The detail calculation of InGaAs QD is described by R. Melnik et al [22]. In our simulation, the bowing factor of calculating the band gap used as 1.4 eVand the band offset ratio was assumed to be 60 : 40 [23-24]. As a result the potential of conduction and valence band is defined as

0.6 And the effective mass of InGaN is used by the linear combination of InN and GaN.

1

The simulation results of InGaN/GaN QD structure are shown in Figure 2.3.2 (a), (b) and (c). Those figures demonstrate electron wavefuction in ground states, first excited and second excited states. The existing probability of carrier is found to confine in the QD with lower potential than barrier region. One the other hand, the corresponding eigenvalue of electron or hole can be obtained in the software.

As the QD size was increasing, the energy levels of quantized sub-band become smaller. The relation was shown in Figure 2.3.3 when we consider the InGaN QD with 3 nm and 10 nm in diameter and height. The quantum confinement energy was almost proportional to L^-2, where L was the potential dimension. And the build-in fields under electric field 0, 0.25, 0.5 M Vol/cm² are considered. The build-in field is found to result in the reduction of resonance energy. The wavefution of electon attend to exist in one side of bending potential. Under larger built-in field, the obvious

dependence on the well width can be recognized in the figure 2.3.3.

2.4 The localization effect in quantum well structure

In InGaN-based structure, the high luminescence efficiency is due to In-rich regions within InGaN layer. These In-rich regions are believed to act as In-rich QDs in the InGaN layer, providing deep potential wells that suppress the diffusion of electrical carriers toward various nonradiative defects and result in a large red shift in the emission energy.

The behavior of In-rich or QDs like region in InGaN QWs are a unique characteristic and the origins of Indium segregation are not very clear until now.

There are several suggestions to explain the production of In rich cluster and relative QD-like structure. One of the possibilities is resulted from the compositional fluctuation. The compositional fluctuation of indium is commonly observed in the majority of InGaN alloys. Another possibility is resulted from phase separation.

Theoretical calculations predict that phase separation in an InGaN layer occurs below the critical temperature and for a range of composition of the alloy that defines a miscibility gap at a given growth temperature. Otherwise, in the experimental results,

In-rich QDs formed by phase separation have been observed in InGaN films that contain high concentrations of In or a layer thickness larger than the critical layer thickness. Moreover, InGaN MQW always has a so-called V-defect consist of treading dislocation. It has been proposed that the V-defect was proposed to attach with and associated with In-rich regions. Cremades et al [24] suggested that the built in strain in InGaN layers determines the preferential incorporation of In atoms to the ally in the regions where the strain is relax thus resulted in the formation of defect.

2.5 The basic concept of nonpolar stucture

Wurzite GaN-based structure grown in nonpolar orientation is attracting attention owing to less influence of the built in field in QW. The built in field is the sum of spontaneous polarization and strain induced piezoelectric field. In this section, the basic concept of nonpolar structure will be discussed. The structure diagram of hexagonal structure has been demonstrated in Fig 1.1.3. In normal, nonpolar structure can be developed on two growth method. The first approach is growing [1100]

oriented m-plane GaN template on closely lattice matched [100] plane of LiAlO2 by hydride vapor phase epitaxy and molecular beam epitaxy [7]. The Second is growing [1120] oriented a-plane GaN template on r-plane sapphire or [1120] SiC by metalorganic vapor-phase (MOVPE) [8]. These two methods has been successfully developed on AlGaN/GaN and InGaN/GaN MQW structure [26-27]. Based on conventional structure, the build in field is supplied by strain induced piezoelectric field and spontaneous polarization field. The build in field of GaN based structure is induced from Ga and N atoms which arranged parallel growth axis in normal c-plane films. The piezoelectric and spontaneous electric fields are at the opposite direction as shown in Figure 2.5.1. However, nonpolar structure was applied to enhance the internal quantum efficiency due to not affected by polarization effect based on the direction of Ga and N atoms arrange normal to the growth axis. As a result, the build in electric field is normal to the direction of QW. The potential can not be influenced by the polarization and its shape keep on flat as shown in Fig 2.5.2 The wavefuciton overlap of electron and hole is larger than c-plane GaN.

Fig 2.1.1 The diagram of strain relaxation for S-K growth mode

Fig 2.1.2 A SEM top-view and cross-sectional view (b) of InGaN QD structures

x 1

Table 2.2.1 Binding energy and relative simulation factors of III-V semiconductors

z

Fig.2.3.1 (a) Wavefunction of electron in ground state Max 5.615

Min - 0.0347 InGaN

GaN

r

Max 11.238

Min - 0.011

Fig.2.3.1 (b) Wavefunction of electron in first excited state

Max 6.753

Min - 2.768

Fig.2.3.1(c) Wavefunction of electron in second excited state

1 2 3 4 5 -150

-100 -50 0 50 100 150 200 250 300 350

Fz=0 M volt/cm² FZ=0.25 M volt/cm² F_z=0.5 M volt/cm²

Resonance energy (eV)

The heigth of QD (nm)

Fig.2.3.2 Well width dependence of the ground state resonance energies in the single QD model under electric field 0, 0.25, 0.5 M Vol/cm²

Fig.2.5.1 Band diagram of c-plane InGaN/GaN MQW

Fig.2.5.2 Band diagram of a-plane InGaN/GaN MQW

Chapter 3 Experimental principle and Experiment Setup

Photoluminescence is the emission of light from a material under optical excitation.

The energy of the laser light should be shorter than the band gap energy of the semiconductor. The exciting source is absorbed by semiconductor and the electros in conduction band and holes in valance band generated under excitation. When an excited electron in an excited state returns to initial state forming a photon a photon whose energy is the difference between the excited state and the initial state energies, then we detect the PL signal. In order to reach equilibrium, the carriers will recombination in many ways. The processes can be direct or indirect depending on the band gap energy of the material. The PL spectra are obtained by analyzing the spectral elements of the emitted light. Depending on the type of the band gap the transition can be classified into direct transition and indirect transition. The combination of indirect transition related to the phonon scattering which results in the momentum and the energy transition.

Typical radiative recombination processes in semiconductors occur in many ways, PL emission of those transitions can be recognized easily at low temperature exclusive of the thermal energy. The exciton was generated by of electron and hole pair which bounded by the coulomb interaction. Luminescence of high purity and high quality semiconductors is dominated by free exciton (FE) emission, normally refereed to a Wannier-Mott exciton [28]. The FE emissions from h-GaN at low temperature have been observed in the early 1970s. There are many forms of exciton bounded to other particles. An exciton bound to a neutral donor and is in general called a bound excton (BE). Similarly, neutral acceptors always produce exciton bound to them. The transition energy of BE is lower the binding energy than that of FEs [29]. The simultaneously existence of doner and acceptor impurities introduce a

pair called doner-acceptor pair (DAP) recombination.

The setup of our PL system is shown in Figure 3.3.1. The pumping light source was multi-mode and non-polarized Helium-Cadmium laser operated on 325nm with 20mW. After reflected by three mirrors, the laser light was focus by a lens which focal length was 5cm, to 0.1 mm in diameter and the luminescence signal was collected by some lens. The probed light was dispersed by 0.32 monochromator (Jobin-Yvon Triax-320) equipped with 1800, 1200, and 300 grooves/mm grating and which maximum width of the entrance slits was 1mm. The resolution was controlled in 1nm by selecting 300 grooves/mm grating and slit of 0.1 mm. We use long pass filter in order to avoid the laser coupling with the PL spectrum.

3.2 Photoluminescence Excitation (PLE)

In PL measurement, which is performed at fixed excitation, is performed at fixed excitation energy, the luminescence properties are generally investigated. While PL excitation (PLE) spectroscopy, which is carried at fixed detection energy, provides mainly information about the absorption properties. Apart from absorption and PL experiment is widely used spectroscopic tool for the characterization of optical transitions in semiconductors. It is very important to note that the PLE also depends strongly on the different carrier relaxation processes. Nevertheless, in many cases it is difficult to separate the influence of relaxation from that of absorption. The PLE spectrum is strongly influenced by the relaxation depending on different samples.

The setup of PLE is shown in Figure 3.3.2.The pumping source of PLE was Xe lamp with 450W separated by double-grating monochoramator ( Jobin-Yvon Gemini 180) and then coupled to samples at an angle about 45˚ by two focal lenses. We fixed the detection energy of the spectrometer Triax 320, and changed the excitation energy of Xe lamp from 300 nm to the band gap of each sample. At the exit of the

spectrometer Triax 320, a high sensitive Hamamatsu photomultiplier tube (PMT) with GaAs photocathode was placed to detect the luminescence signals.

3.3 Atomic Force Microscopy

Atomic force microscopy (AFM) images the surface of a sample by scanning a

Atomic force microscopy (AFM) images the surface of a sample by scanning a