Transport imaging is a SEM-based technique used to directly image the motion and recombination of charge in luminescent semiconductors, allowing for the extraction of transport parameters critical to device operation. In this thesis, transport imaging for 1D structures was initiated with work on sample preparation, modeling and initial characterization. One dimensional structures are being integrated into forefront electronics due to their inherent advantages in size, packing density and power consumption. In this work the one dimensional equation for steady state minority carrier recombination distribution solved for the Gaussian source is derived and results from numerical simulations are presented. The diameter of the SEM beam is determined experimentally allowing for accurate simulation parameters. Intensity and drift measurements on four batches of top-down wire structure samples, fabricated on a AlGaAs/GaAs/AlGaAs double heterostructure using a FIB, are presented. Significant decreases in luminescence in FIB exposed regions are reported. Spatial luminescence from single bottom-up GaN and ZnO nanowires deposited by metal initiated metal-organic CVD on Au and SiO2 substrates is imaged. CL spectra for GaN and ZnO, with peak intensities at 3.27 and 3.29 eV, are characterized. Finally, several suggestions for further research are offered including transport imaging on contacted bottom-up nanowires and a potential application of transport imaging to FIB damage characterization.

Master of Science / Department of Electrical and Computer Engineering / Behrooz Mirafzal / Wide bandgap devices have gained increasing attention in the market of power electronics for their ability to perform even in harsh environments. The high voltage blocking and high temperature withstanding capabilities make them outperform existing Silicon devices. They are expected to find places in future traction systems, electric vehicles, LED lightning and renewable energy engineering systems. In spite of several other advantages later mentioned in this paper, WBG devices also face a few challenges which need to be addressed before they can be applied in large scale in industries. Electromagnetic interference and new requirements in packaging methods are some of the challenges being faced by WBG devices. After the commercialization of these devices, many experiments are being carried out to understand and validate their abilities and drawbacks. This paper summarizes the experimental results of various applications of mainly Silicon Carbide (SiC) and Gallium Nitride (GaN) power devices and also includes a section explaining the current challenges for their employment and improvements being made to overcome them.

Gallium Nitride (GaN) power devices have the potential to jump-start the next generation of power converters which are smaller, faster, denser, and cheaper. They are thus expected to meet the increasing 21st Century need for power density and efficiency, while at the same time reducing pollution. With the commercialisation of 600 V GaN power devices, which the industry is keen to adopt, come significant challenges. Since there are a number of such devices which are new to the power community, there is a steep learning curve involved, with dispersed information on how best to employ these devices. This work aims to solve this problem through the development of a universal GaN power device and circuit model and the formulation of design rules and guidelines. Through this contribution, designers will be able to better understand and work with these novel devices with relative ease. This will aid the need for faster adoption of GaN devices by the industry solving the barriers to commercialisation. This research demonstrates the use of machine learning (ML) algorithms for behavioural modelling of GaN power devices. Introducing ML as the key to developing a general behavioural and circuit model for GaN power devices combined with understanding, learning, customizing and successfully demonstrating it is the major contribution of this research work. This research first presents a comprehensive investigation into the parasitic effect on the GaN device switching performance. A simple process based on RF techniques is introduced to approximately extract the impedances of the GaN device to develop a behavioural model. The switching behaviour of the model is validated using simulation and double pulse test experiments at 450 V, 10 A test conditions. The developed behavioural model for Transhporm GaN HEMT is 95.2% accurate as the existing LT-spice manufacturer model, and is very much easier for power designers to handle, without the need for knowledge about the physics or geometry of the device. However, given that separate models would need to be developed for each commercial GaN device, the need for a generalized and accurate GaN behavioural model was identified, and it is this generalised model that the remainder of this thesis focuses on. In the next part of this research, a GaN platform test bench is built through bridging RF and power electronics design methodologies to achieve a gate loop and power loop inductance of around 1.8nH with switching waveforms with rise time and fall time around 2.5ns at 450V, 15A, 500KHz test conditions. The double pulse test circuits are customized using different off the shelf gate drives and analysed for collecting switching data for training the ML model. ML modelling using supervised learning is used to predict the switching voltage and current waveforms thus making it possible to construct a generic GaN black box model. Different architectures with single and multi- layer neural networks are explored for modelling. The ability to demonstrate a GaN device ML model that maps both voltage and current inputs and outputs is another characteristic and novel feature of this work. This research demonstrates different types of GaN ML models. The developed voltage and current prediction models are based on feed forward neural network (FFNN), long short-term memory unit (LSTM) and gated recurrent unit (GRU). Several parameters are quantified and compared for validating the models. They are the network architectures, parameters, training time, validation loss and error loss. The ML models are also compared with the demonstrated model of chapter 3 and existing LT-Spice manufacturer models. The results show that the author has been able to develop a GaN LSTM ML model with an error rate of 0.03, and convergence at 3s with excellent stability. The ML based modelling is then translated from GaN power devices to GaN based circuits. Among the different neural network architectures trained and tested, a multi FFNN with 5 hidden layers and 30 neurons, was found to be the best for prediction and optimization. The switching behaviour comparison results shows the benefits and value of ML modelling in opening up whole new possibilities of employing advanced control algorithms for very efficient, reliable and scalable performance of GaN power electronics systems. Finally, the findings of this work have been generalized to frame machine learning based techniques to address the need for generic modelling of power electronic devices. These solutions are presented as an information manual to researchers, engineers and students interested in benefiting from adopting machine learning for power electronics applications.

In this thesis, optical studies of c-plane InGaN/GaN quantum well (QW) structures are presented. The effects of a Si-doped underlayer (UL) on the optical properties of multiple quantum well (MQW) structures are investigated. The QW photoluminescence (PL) emission peak energy and radiative recombination rate decrease and increase respectively with increasing number of QWs. These observations are attributed to the increasing net electric field across the MQW structure as the strength of the surface polarisation field, which acts in the opposite sense to the piezoelectric polarisation fields across the QWs, reduces with increasing distance of the UL from the sample surface. This leads to a reduction in the electron-hole recombination energy and wavefunction overlap. It is also shown that the internal quantum efficiency of the MQW structures may decrease with increasing number of QWs due to the reducing radiative recombination rate, which could indicate that carrier losses due to thermionic emission or interface recombination are mitigated by the inclusion of an UL. Optical studies of single QW structures containing Si-doped ULs with different net electric fields across the QW are presented. The net electric field across the QW is changed by varying the thickness of the GaN cap layer. The full width at half maximum of the emission peak increases with increasing net electric field across the QW. This is attributed to the increasing variation in electron ground state energies due to the role of the electric field in the localisation of electrons at quantum well width fluctuations. For one sample, a smaller Huang-Rhys factor compared to the rest of the samples is calculated. The non-exponential PL decays detected on the low energy side of the QW emission peak from this sample are also of a different shape to the other PL decays detected at all energies for the other samples. This may be due to the reversal of the net electric field across these QW regions. Observations of a broad emission band on the high energy side of single QW structures at high excited carrier densities are presented. This band occurs in the carrier density regime at which the efficiency droop is observed. The emission band is attributed to higher energy weakly localised or delocalised electron and hole states that are populated following the saturation of the localised ground states. PL decay curves detected across this emission band exhibit plateaus where the PL intensity remains constant until the higher energy emission has decayed. These are similar to decays observed in semiconductor quantum dots, which are characteristic of Pauli state blocking.

In this thesis, I report on the investigation of the low frequency vibrational modes in a number of different systems using ultrafast spectroscopic techniques. These consist of biological systems, including the enzyme, morphinone reductase (MR) and the related biomolecules, riboflavin (Rb) and flavin mononucleotide (FMN), as well as non-biological systems, including the semiconductor gallium nitride (GaN) and gold nanoparticles (Au NPs). The term low frequency refers to terahertz (THz) frequencies, where vibrational modes exist at the molecular level, with molecular rotations, lattice vibrations and inter- and intra-molecular vibrations occurring in the THz spectral range. These vibrational modes occur on sub-picosecond timescales and therefore ultrafast techniques utilising femtosecond laser pulses provide a means of studying these modes, and are employed throughout this thesis. The two ultrafast techniques of transient absorption (TA) spectroscopy and terahertz time-domain spectroscopy (THz-TDS) were used. Firstly, a high-repetition rate transient absorption (HRRTA) spectrometer was commissioned to perform pump-probe measurements with an ultraviolet pump and broadband visible probe. The performance of the HRRTA spectrometer was benchmarked using Au NPs and used to investigate the existence of a promoting vibration in MR contributing to the catalysis process, predicted to occur at THz frequencies. Weak oscillations were detected in the charge-transfer absorption band of MR bound to the non-reactive cofactor 1,4,5,6-tetra-hydro-nicotinamide adenine dinucleotide (NADH4), with a frequency of approximately 1.5 THz and provide evidence of the first direct observation of a promoting vibration in an enzyme. To complement the TA measurements, THz-TDS was also used to obtain direct measurements of the absorption at THz frequencies. Due to the challenge of studying water-based biological samples, an initial investigation was performed on a wurtzite GaN wafer, which exhibited optical phonon modes in the THz frequency range that were found to determine the dielectric response of the semi-insulating semiconductor wafer. Use of a non-polar m-plane wafer allowed the anisotropic nature to be observed and values of 9.22 ± 0.02 and 10.32 ± 0.03 for the static dielectric constants were obtained for the THz electric field polarised both perpendicular and parallel to the c-axis of the wurtzite GaN wafer, respectively. Finally, biological studies using THz-TDS were performed with measurements on Rb pellets and films revealing vibrational modes in the THz region. The sharp absorption features were not observed in FMN, despite a small difference in molecular content from Rb, and dehydration was required to reveal small amplitude absorption features. Final measurements on MR and MR-NADH4 films were carried out and evidence of absorption features in the THz frequency range were observed, however further work is required to determine the precise origin of these features.

A thesis submitted to the Faculty of Science, University of the Witwatersrand, Johannesburg, in fulfilment of the academic requirements for the degree of Doctor of Philosophy. Johannesburg, 2014. / Unable to load abstract.

<p><p>In this work growth of cubic GaN in the selective area (SA) MOVPE process is</p><p>simulated. The simulations are restricted to small pattern SA MOVPE growth.</p><p>In this case the traditional MOVPE growth and the enhanced growth caused by</p><p>surface diffusion are important growth factors. The lateral vapor phase diffusion</p><p>is ignored while this process only has a small impact on the enhanced growth in</p><p>the small pattern SA growth. The model is build for simulation of anisotropic</p><p>growth. It has been shown that different type of anisotropic growth occurs when</p><p>the mask pattern are orientated in different directions on the substrate. While</p><p>the anisotropic growth is not well understood two different models are studied in</p><p>this work.</p><p>The simulation is restricted to the geometrical growth characteristics such</p><p>as mask and crystal width, mask alignment and surface diffusion on the crystal.</p><p>The reactor geometry, pressure and growth temperature are not investigated that</p><p>closely and are only treated as constants in the model.</p><p>The model used in this simulation gives good results for short time simulations</p><p>for some certain cases. The model shows that the fill factor has a greater</p><p>impact on the grown shapes than the individual mask and crystal width. But</p><p>there are problems with the anisotropic and flux from mask modeling while some</p><p>facets do not appear and the lateral growth along the mask show doubtful results.</p><p>The model show good results in short time growth and predict some important</p><p>characteristics in SA MOVPE.</p></p>

Uniformly distributed ZnO nanorods with diameter 70-100 nm and 1-2μm long have been successfully grown at low temperatures on GaN by using the inexpensive aqueous solution method. The formation of the ZnO nanorods and the growth parameters are controlled by reactant concentration, temperature and pH. No catalyst is required. The XRD studies show that the ZnO nanorods are single crystals and that they grow along the c axis of the crystal plane. The room temperature photoluminescence measurements have shown ultraviolet peaks at 388nm with high intensity, which are comparable to those found in high quality ZnO films. The mechanism of the nanorod growth in the aqueous solution is proposed. The dependence of the ZnO nanorods on the growth parameters was also investigated. While changing the growth temperature from 60°C to 150°C, the morphology of the ZnO nanorods changed from sharp tip (needle shape) to flat tip (rod shape). These kinds of structure are useful in laser and field emission application. / Singapore-MIT Alliance (SMA)

High density, uniform GaN nanodot arrays with controllable size have been synthesized by using template-assisted selective growth. The GaN nanodots with average diameter 40nm, 80nm and 120nm were selectively grown by metalorganic chemical vapor deposition (MOCVD) on a nano-patterned SiO2/GaN template. The nanoporous SiO2 on GaN surface was created by inductively coupled plasma etching (ICP) using anodic aluminum oxide (AAO) template as a mask. This selective regrowth results in highly crystalline GaN nanodots confirmed by high resolution transmission electron microscopy. The narrow size distribution and uniform spatial position of the nanoscale dots offer potential advantages over self-assembled dots grown by the Stranski–Krastanow mode. / Singapore-MIT Alliance (SMA)

In this work growth of cubic GaN in the selective area (SA) MOVPE process is simulated. The simulations are restricted to small pattern SA MOVPE growth. In this case the traditional MOVPE growth and the enhanced growth caused by surface diffusion are important growth factors. The lateral vapor phase diffusion is ignored while this process only has a small impact on the enhanced growth in the small pattern SA growth. The model is build for simulation of anisotropic growth. It has been shown that different type of anisotropic growth occurs when the mask pattern are orientated in different directions on the substrate. While the anisotropic growth is not well understood two different models are studied in this work. The simulation is restricted to the geometrical growth characteristics such as mask and crystal width, mask alignment and surface diffusion on the crystal. The reactor geometry, pressure and growth temperature are not investigated that closely and are only treated as constants in the model. The model used in this simulation gives good results for short time simulations for some certain cases. The model shows that the fill factor has a greater impact on the grown shapes than the individual mask and crystal width. But there are problems with the anisotropic and flux from mask modeling while some facets do not appear and the lateral growth along the mask show doubtful results. The model show good results in short time growth and predict some important characteristics in SA MOVPE.