PhD Dissertations

Permanent URI for this collection

Browse

Recent Submissions

Now showing 1 - 5 of 1159
  • Dissertation

    Artificial-intelligence driven design of liquid fuels.

    (2024-03) Kuzhagaliyeva, Nursulu; Sarathy, S. Mani; Ghanem, Bernard S.; Grande, Carlos A.; Green, William H.; Physical Science and Engineering (PSE) Division

    Developing high-performance fuels from low-carbon or carbon-neutral streams can significantly advance decarbonization efforts in the transportation sector. Given that practical fuels are complex mixtures of hundreds of components, and considering the highly non-linear nature of fuel properties, the task of exploring a broad chemical space for potential blendstocks becomes a cyclical, costly, and time-consuming process. In the era of matter engineering, inverse design has become a vital component in the sophisticated framework required for the rapid design of fuels. The swift advancements in artificial intelligence, particularly in machine learning, have propelled the application of inverse design in various fields, where specific properties are pre-selected to identify suitable new candidates. This dissertation introduces an inverse, data-driven design framework for liquid fuel formulation that employs a constrained optimization approach. Central to this framework are predictive deep learning (DL) models and robust search algorithms for efficient chemical space navigation. A novel contribution of this work is the integration of an algorithmic advancement into the training loop, directly connecting molecular structures with mixture representations and facilitating mixture-level fuel screening. This level of screening evaluates several crucial combustion-related properties, including octane rating, sooting propensity, and volatility, as well as the emissions of unburnt hydrocarbons (UHCs). The functionality of the proposed framework was extended to screen UHC formation trends across a diverse range of fuels, taking into account the effects of fuel composition and operating conditions (\ie pressure, temperature, and equivalence ratio). To effectively screen the quantities of important intermediates and by-products in complex gas-phase reactive systems, we adopted a generic one-dimensional burner-stabilized stagnation flame model using state-of-the-art kinetic mechanisms. The framework's effectiveness is demonstrated through the design of high-octane, low-sooting fuels that adhere to gasoline specification constraints, using a variety of gasoline blendstocks. This work also assesses various molecular representations and state-of-the-art deep learning practices. Finally, the dissertation presents a search workflow through several case studies that directly optimize mixtures within the domains of fossil, oxygenated, and synthetic streams. We expect our simple and practical framework will serve as a solid baseline and help ease future research designing liquid energy carriers.

  • Dissertation

    Enhancing the Bioelectronic Interface through Material and Device Engineering

    (2024-01-24) Saleh, Abdulelah; Inal, Sahika; King Abdullah University of Science and Technology (KAUST); Adamo, Antonio; Li, Mo; Owens, Roisin; Biological and Environmental Science and Engineering (BESE) Division

    The integration of bioelectronic devices in medical and environmental monitoring systems and their use as therapeutic agents have made significant strides in recent years, propelled by advancements in material science and device engineering. This thesis includes bioelectronic sensors and actuators developed for communicating with various biological interfaces, such as the human skin, in vitro cell culture, and physiological media. The common focus of these distinct devices is the bioelectronic interface enhanced through the selection of the appropriate material, device components, and configurations, to achieve a seamless integration between electronic devices and biological systems. The first contribution of this work includes the development of flexible and conformable electrodes based on inkjet-printed MXene films. These films are formed from aqueous solutions on soft and self-standing polymeric substrates which generate a conformable interface with the skin. The device is demonstrated for cutaneous biosensing applications, particularly for electrocardiogram (ECG) monitoring and sweat analysis. I highlight the importance of selecting mechanically compliant system elements to detect signals from the human skin and sweat that is produced therein. I next focus on the type of devices that can be used to improve signal transmission from biological fluids. I tackle the question of whether an electrode or a transistor configuration leads to higher performance in sensing metabolites using conducting polymer films functionalized with a redox enzyme. I compare the performance of an organic electrochemical transistor (OECT) configuration with conventional electrode setups, revealing the tradeoffs of each system in terms of sensitivity and dynamic range. The next interface is the one between in vitro stem cell cultures and electrical stimulation devices, which are designed to modulate their differentiation characteristics. I evaluate the efficiency of bioreactor configurations with different electronic materials for optimal electrical pacing of human pluripotent stem cell-derived cardiomyocytes (hPSC-CMs) and show the importance of electrode material and configuration for maximized stimulation performance. In addition to investigations on the effect of active electrical stimulation on the behavior of these cells, I discover that conducting polymer coatings underneath the cells can also interfere with their functionality. I show that the same cells enhance their maturation properties depending on the substrate material used, highlighting the role of a passive interface in governing the faith of in vitro tissues. Although each interface I focused on with a new device is unique, lessons learned from one interface can translate to another. Overall, this dissertation provides a comprehensive study on the enhancement of bioelectronic interfaces by focusing on the electronic material, configuration, and substrate selection, aiming to provide guidelines on bioelectronic device design.

  • Dissertation

    Interface engineering for GaN HEMTs

    (2024-02-26) Wang, ChuanJu; Li, Xiaohang; King Abdullah University of Science and Technology (KAUST); Salama, Khaled Nabil; Zhang, Xixiang; Zhao, Yuji; Computer, Electrical and Mathematical Science and Engineering (CEMSE) Division

    The junctions formed between metallic gate electrodes and GaN, gate dielectrics and GaN are crucial components of GaN-based electronics. Schottky gate contact formed between metallic film and GaN has great importance in controlling the performance of GaN-based high-electron-mobility transistors (HEMTs). However, interface defects generate at the metal-GaN interface can alter the electrical characteristics of the GaN HEMTs, influencing parameters such as threshold voltage, ION/IOFF current, subthreshold swing, and channel carrier mobility. On the other hand, at the dielectric-GaN interface, insulators such as Al2O3, SiO2, and SiN with high dielectric constant and wide bandgap are used as gate dielectrics to suppress the gate leakage current. However, numerous traps and fixed charges are introduced at the dielectric/GaN interface, deteriorating the device reliability and shifting the threshold voltage. The attempt to achieve a high quality interface with low interface traps is the main part of GaN HEMTs study.
    The first part of this thesis concentrates on the metal-GaN interface, we found the e-beam evaporation process creates considerable damage to the GaN surface. The induced interface traps and rough oxide at the Ni/Au-GaN interface can grossly degrade the performance of GaN HEMTs. To protect the GaN surface, Ti3C2Tx MXene films were spray-coated onto the AlGaN/GaN epitaxial wafer as the gate contact. The van der Waals heterojunction between MXene films and GaN without direct chemical bonding retained the pristine atomically flat surface of GaN, a record high ION/IOFF current of 1013 and an ultra-low subthreshold swing of 61 mV/dec was obtained. At the gate dielectric/GaN interface, we explored the Al2O3/Si and Al2O3/GaN heterostructures by comparison study to clarify the origin of the interfacial charges. Additionally, we found dipole forms at the interface due to electron transfer and redistribution at the Al2O3 /GaN interface. Dipole induced electric field at the Al2O3/GaN interface can shift the threshold voltage to a negative value. At last, we studied the Al2O3/GaN interface, we have grown a thin Ga2O3 interlayer by the pulsed laser deposition technique between Al2O3 and GaN. The resulting interface traps was significantly suppressed compared with the Al2O3/GaN capacitor.

  • Dissertation

    Modeling of complex turbulent and multiphase flows for sustainable power production

    (2023-12) Ceschin, Alberto; Im, Hong G.; King Abdullah University of Science and Technology (KAUST); Cuoci, Alberto; Castaño, Pedro; Roberts, William L.; Sun, Shuyu; Physical Science and Engineering (PSE) Division

    As industrialization and fossil fuels have dominated the power sector for decades, electricity generation is one of the most significant anthropogenic carbon dioxide (CO2) releasing activities. Climate awareness has grown dramatically throughout the world, with ambitious goals such as an emission-free power sector by 2035 and a net-zero carbon economy by 2050 in the United States, an emission peak by 2030 and carbon neutrality by 2060 in China, and net-zero carbon by 2070 in India \cite{cozzi2020world}. To meet these aims, existing fossil fuel power plants must reduce CO2 emissions to zero or near-zero levels quickly and deeply. Nonetheless, fossil fuel power plants constitute CO2-emitting infrastructure with high socio-technical inertia and, as of today, no carbon-free alternative stands out. Therefore, a shift will likely be immediate and abrupt: long-distance net-zero targets have to be coupled with rapid mitigation of the environmental impact. Hence the present work. In a multifaceted research world, Computational fluid dynamics (CFD), in the nuance of Finite Volume (FV), was chosen as the Swiss knife to secure short and medium-distance targets. In detail, the present thesis aims to describe turbulent and multiphase flow for sustainable energy applications. The common tool is the OpenFOAM library, chosen for its high flexibility as a bridge between theoretical fundamentals and industrial applications. The first part of the thesis deals with fundamental aspects regarding present technologies related to power generation. They are in order: the nature of the high level of turbulence in a well-studied premixed burner, the break up of a viscous vacuum residue oil (VRO) jet in a gasifier and the liquid injection in a compression ignition engine are the subjects of the study. On the other side, the second part focuses on novel and promising technologies, such as Ultrasonically Induced Cavitation (UIC) for oil conversion and Cryogenic Carbon Capture (CCC) for carbon dioxide removal.

  • Dissertation

    Natural Language Models for Unnatural Languages

    (2023-11-20) Para, Wamiq Reyaz; Wonka, Peter; King Abdullah University of Science and Technology (KAUST); Rohrbach, Anna; Tegner, Jesper; Hadwiger, Markus; Computer, Electrical and Mathematical Science and Engineering (CEMSE) Division

    Generative models for language generation, particularly based on transformers have shown remarkable performance in domains dealing with language. More recent works have attempted to repurpose these models to perform image generation from text prompts leading to works like DALL-E, which can generate realistic images. This dissertation aims to describe our framework that adapts generative models of languages, allowing them to generate non-language objects. This proposed framework allows for generation in a common framework across seemingly unrelated domains. We tackle three problems of increasing complexity - room- layout generation, CAD-sketch generation, and furniture-layout generation. We solve each of the problems by creating a grammar that describes each object in the problem domain representing each object by a sequence of tokens. We then train a transformer model over these sequences. Sampling from this transformer is equivalent to sampling the object that the generated sequence represents. In room-layout generation, we show how to represent a room-layout as a sequence. In CAD-sketch generation, we create a more complex grammar that is not context-free and show how to parse this complex grammar. In furniture-layout generation, we describe a model that subsumes the functionality of both the previous models and adds a new property - the model is not restricted to a left-to-right generation order that most generative language models possess.