Machine learning is a branch of artificial intelligence that uses data to automatically build inferences and models designed to generalise and make predictions. In this thesis, the use of machine learning in materials science is explored, for two different problems: the optimisation of gallium nitride optoelectronic devices, and the prediction of material failure in the setting of laboratory earthquakes. Light emitting diodes based on III-nitrides quantum wells have become ubiquitous as a light source, owing to their direct band-gap that covers UV, visible and infra-red light, and their very high quantum efficiency. This efficiency originates from most electronic transitions across the band-gap leading to the emission of a photon. At high currents however this efficiency sharply drops. In chapters 3 and 4 simulations are shown to provide an explanation for experimental results, shedding a new light on this drop of efficiency. Chapter 3 provides a simple and yet accurate model that explains the experimentally observed beneficial effect that silicon doping has on light emitting diodes. Chapter 4 provides a model for the experimentally observed detrimental effect that certain V-shaped defects have on light emitting diodes. These results pave the way for the association of simulations to detailed multi-microscopy. In the following chapters 5 to 7, it is shown that machine learning can leverage the use of device simulations, by replacing in a targeted and efficient way the very labour intensive tasks of making sure the numerical parameters of the simulations lead to convergence, and that the physical parameters reproduce experimental results. It is then shown that machine learning coupled with simulations can find optimal light emitting diodes structures, that have a greatly enhanced theoretical efficiency. These results demonstrate the power of machine learning for leveraging and automatising the exploration of device structures in simulations. Material failure is a very broad problem encountered in a variety of fields, ranging from engineering to Earth sciences. The phenomenon stems from complex and multi-scale physics, and failure experiments can provide a wealth of data that can be exploited by machine learning. In chapter 8 it is shown that by recording the acoustic waves emitted during the failure of a laboratory fault, an accurate predictive model can be built. The machine learning algorithm that is used retains the link with the physics of the experiment, and a new signal is thus discovered in the sound emitted by the fault. This new signal announces an upcoming laboratory earthquake, and is a signature of the stress state of the material. These results show that machine learning can help discover new signals in experiments where the amount of data is very large, and demonstrate a new method for the prediction of material failure.

The increased use of DC-consuming electronics in many applications relevant to everyday life, necessitates significant improvements to power conversion and distribution methodologies. The surge in mobile electronics created a new power application space where high efficiency, size, and reduced complexity are critical, and at the same time, many computational tasks are relegated to centralized cloud computing centers, which consume significant amounts of energy. In both those application spaces, conversion and distribution efficiency improvements of even a few-% proves to be more and more challenging. A lot of research and development efforts target each source of loss, in an attempt to improve power electronics such that it serves the advances in other fields of electronics. Non-isolated DC-DC converters are essential in every electronics system, and improvements to efficiency, volume, weight and cost are of utmost interest. In particular, increasing the operation frequency and the conversion ratio of such converters serves the purposes of reducing the number or required conversion steps, reducing converter size, and increasing efficiency. The aforementioned improvements can be achieved by using superior technologies for the components of the converter, and by implementing higher level of integration than most present-day converters exhibit. In this work, Gallium Nitride (GaN) high electron mobility transistors (HEMT) are utilized as switches in a half-bridge buck converter topology, in conjunction with fine-line 180nm complementary metal oxide semiconductor (CMOS) driver circuitry. The circuits are integrated through a face-to-face bonding technique which results in significant reduction in interconnects parasitics and allows faster, more efficient operation. This work shows that the use of GaN transistors for the converter gives an efficiency headroom that allow pairing converters with state-of-the-art thin-film inductors with magnetic material, a task that is currently usually relegated to air-core inductors. In addition, a new "core-clad" structure for thin-film magnetic integrated inductors is presented for the use with fully integrated voltage regulators (IVRs). The core-clad topology combines aspects from the two popular inductor topologies (solenoid and cladded) to achieve higher inductance density and improved high frequency performance.

by Lam Siu Dan. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2001. / Includes bibliographical references (leaves 65-67). / Text in English; abstracts in English and Chinese. / by Lam Siu Dan. / Table of contents --- p.I / Chapter Chapter 1 --- Introduction / Chapter 1.1 --- Interest in GaAs1-xNx/GaAs alloy --- p.1 / Chapter 1.2 --- Interest in growing GaAs1-xNx/GaAs using different carrier gases --- p.4 / Chapter 1.3 --- Theoretical calculation of the band gap energy of GaAs1-xNx alloy --- p.4 / Chapter 1.4 --- Advantages of using photoluminescence (PL) --- p.7 / Chapter 1.5 --- Our work --- p.9 / Chapter Chapter 2 --- Experimental setup and procedures / Chapter 2.1 --- Growth conditions of GaAs1-xNx on (001) GaAs --- p.10 / Chapter 2.2 --- X-ray diffraction / Chapter 2.2.1 --- Setup --- p.12 / Chapter 2.2.2 --- Types of X-ray measurements --- p.12 / Chapter 2.3 --- PL measurements / Chapter 2.3.1 --- Setup --- p.14 / Chapter 2.3.2 --- Types of PL measurement --- p.16 / Chapter Chapter 3 --- Results and discussions / Chapter 3.1 --- X-ray diffraction of GaAs1-xNx/GaAs / Chapter 3.1.1 --- GaAs1-xNx/GaAs grown using H2 as carrier gas --- p.17 / Chapter 3.1.2 --- GaAs1-xNx/GaAs grown using N2 as carrier gas --- p.28 / Chapter 3.1.3 --- Peak widths of the X-ray rocking curves of GaAs1-xNx/GaAs --- p.30 / Chapter 3.2 --- Room temperature (RT) and 10K PL of GaAs1-xNx/GaAs / Chapter 3.2.1 --- The energy of the NBE peak of GaAs1-xNx/GaAs --- p.32 / Chapter 3.2.2 --- The width of the NBE peak of GaAs1-xNx/GaAs --- p.44 / Chapter 3.3 --- Excitation power density (EPD) dependent PL studies of GaAs1-xNx/GaAs / Chapter 3.3.1 --- The energy of the NBE peak of GaAs1-xNx/GaAs --- p.49 / Chapter 3.3.2 --- The width of the NBE peak of GaAs1-xNx/GaAs --- p.55 / Chapter 3.4 --- Temperature dependent PL studies of GaAs1-xNx/GaAs --- p.57 / Chapter Chapter 4 --- Conclusions --- p.62 / References --- p.63

This thesis devoted to the experimental studies of yellow and blue luminescence (YL and BL relatively) bands in Gallium Nitride samples doped with C and Si. The band BLC was at first observed in the steady-state photoluminescence spectrum under high excitation intensities and discerned from BL1 and BL2 bands appearing in the same region of the spectrum. Using the time-resolved photoluminescence spectrum, we were able to determine the shape of the BLC and its position at 2.87 eV. Internal quantum efficiency of the YL band was estimated to be 90\%. The hole capture coefficient of the BLC related state was determined as 7 10-10 cm3/s. Properties of BLC were investigated. The YL and BLC bands are attributed to electron transitions via the (0/-) and (+/0) transition levels of the CN defect.

Nanoporous GaN films are prepared by UV assisted electrochemical etching using HF solution as an electrolyte. To assess the optical quality and morphology of these nanoporous films, micro-photoluminescence (PL), micro-Raman scattering, scanning electron microscopy (SEM), and atomic force microscopy (AFM) techniques have been employed. SEM and AFM measurements revealed an average pore size of about 85-90 nm with a transverse dimension of 70-75 nm. As compared to the as-grown GaN film, the porous layer exhibits a substantial photoluminescence intensity enhancement with a partial relaxation of compressive stress. Such a stress relaxation is further confirmed by the red shifted E₂(TO) phonon peak in the Raman spectrum of porous GaN. / Singapore-MIT Alliance (SMA)

Thesis (Ph. D.)--Ohio State University, 2006. / Title from first page of PDF file. Includes bibliographical references (p. 232-237).

Thesis (Ph. D.)--University of Hong Kong, 2006. / Title proper from title frame. Also available in printed format.

The group III nitride semiconductor grown on c-plane sapphire by radio frequency plasma assisted molecular beam epitaxy has been studied. To archive good quality GaN film, nitridation and low temperature buffer layer were applied to overcome the issue of lattice mismatch. Low temperature and long period nitridation process shows better improved of optical properties and crystal quality of GaN film. Buffer layer grown with slightly Ga-rich, substrate temperature at 522¢J, for 2 minutes leads to better GaN film. High substrate temperature and sufficient nitrogen to gallium ratio are two important factors to control the growth of the good quality GaN epilayer. Chemical etching and observation of surface reconstructions were used to characterize the polarity of group III nitrides. The Ga-polarity GaN film shows 2x surface reconstruction with high chemical resistance while the N-polarity is sensitive to chemical and displays the 3x reconstruction pattern. The process of indium incorporated with GaN is very sensitive to growth temperature. The indium content decreased with increasing the substrate temperature and also decreased along the growth direction. The N-polar GaN with an indium-facilitated growth technique was also studied. Upon the incorporation of indium during growth, the photoluminescence intensity and electron mobility of GaN has been enhanced by a factor of 15 and 6 respectively. The electron concentration drastically increases by several orders of magnitude. The biaxial strain of GaN film estimated with Micro-Raman technique reduces from 0.6729 to 0.5044GPa. The full-widths at half maximum of asymmetric (10-12) x-ray reflection which related to the density of overall threading dislocations increases from 593 to744 arcsec. In contrast, the symmetric (0002) reflection related only to threading dislocations having a non-zero c-component Burgers vectors reduces from 528 to 276 arcsec. The enhancement of GaN optical property is generally attributed to the reduction of non-zero c-component dislocations. The reduction in density is confirmed by cross-sectional transmission electron microscopy.

III-nitrides are a promising material system with unique material properties, which allows them to be utilized in a variety of semiconductor devices. III-nitrides grown by NH3-MOMBE are typically grown with high carbon levels (> 1021 cm-3) as a result of the incomplete surface pyrolysis of the metal-organic sources. Recent research has involved the compensating nature of carbon in III-nitrides to produce semi-insulating films, which can provide low-leakage buffer layers in transistor devices. The aim of this work is to investigate the possibility of forming a 2DEG, which utilizes the highly carbon-doped GaN layers grown by NH3-MOMBE to produce low-leakage buffer layers in the fabrication of HEMTs. These low leakage GaN buffers would provide increased HEMT performance, with better pinch-off, higher breakdown voltages and increased power densities. Additionally, methods of controlling and/or reducing the incorporation of carbon will be undertaken in an attempt to broaden the range of possible device applications for NH3-MOMBE. To realize these transistor devices, optimization and improved understanding of the growth conditions for both GaN and AlGaN will be explored with the ultimate goal of determining the feasibility of III-nitride transistors grown by NH3-MOMBE.

Thesis (Ph. D.)--University of Texas at Austin, 2000. / Vita. Includes bibliographical references (leaves 126-147). Available also in a digital version from Dissertation Abstracts.