136879 research outputs found
Sort by
The Design of a Mobile E-Beam Treatment Station for Contaminated Soils
No full textThe presence of perfluorooctanoic substances (PFAS), a manufactured chemical used in various applications, has raised concerns due to its widespread impact on the environment and hazardous effects on human health. Energetic destructive methods such as thermal, supercritical oxidation, and the electron beam (eBeam) are currently being investigated as possible methods to mitigate this issue. These technologies are more efficient at separating contamination from its media and do not produce a contaminated byproduct. However, these technologies have yet to be proven at a scale at which it can be used to combat this contamination effectively.
This thesis investigates two different design configurations as a scaled pilot for mobile e-beam soil treatment systems by addressing structural and spatial concerns. Structural concerns were addressed by analysis completed in SolidWorks finite element analysis features. The two designs are then compared by considering factors such as radiation shielding, manufacturability, size and throughput, mobility, cost, and ease of setup. PUFFIn+ is used to simulate various electron beam soil interaction parameters such as beam energy, soil thickness and beam utilization efficiency.
These considerations and comparisons result in a final mobile electron beam facility recommendation of the following; the design should be able to be transported using only one trailer, the trailer chassis will require three or more axles and will be an oversize load, the power of the beam should be 3 MeV 50 kW, the optimal thickness of media at this beam energy is 0.5cm, the thickness of shielding if stainless steel is used should be approximately 6 inches around the treatment room, a clamshell shielding should be put around the accelerator, and all conveyors regarding media processing should be metal alloys to avoid contamination. More specific design recommendations are discussed in further detail in the text
Relaxation Dynamics and Thermal Properties of Polyelectrolyte Complexes and Multilayers
No full textPolyelectrolyte complexes and multilayers (PECs and PEMs) are prepared by mixing two oppositely charged polyelectrolytes and by layer-by-layer assembly respectively. PECs and PEMs find applications in drug delivery systems, humidity sensors, electrochemistry, separation membranes, and many more. The key to these applications is understanding the dynamics of polyelectrolyte (PE) chains in PECs and PEMs. The dynamics of PE chain in PECs and PEMs is influenced by temperature, hydration, pH, salt, solvent quality, PE type and, many more. This dissertation discusses two studies on the impact of temperature, water, pH, and, salt on the dynamics of PE chain in PECs and PEMs. Additionally, we also study the influence of salt type on glass transition temperature of PECs.
In the first study, we discuss the impact of salt type on glass transition temperature (Tg) of PECs composed of PSS and poly(diallyldimethylammonium chloride) (PDADMA). We specifically examine the effect of anion type (NaCl, NaBr, NaNO��� and NaI) on Tg. First, we evaluate the effect of salt type on doping of PECs using nuclear magnetic resonance (NMR) and neutron activation analysis (NAA). Next, we evaluate the Tg at different hydration levels using modulated differential scanning calorimetry (MDSC). Put together, these studies give insight into how different parameters such as water, pH, salt and salt type influence the dynamics and relaxation of PE chain in PECs and PEMs.
In the second study, we discuss the effect of temperature, water and pH on relaxation times solid PECs composed of poly(acrylic acid) (PAA) and poly(allylamine hydrochloride) (PAH). It has been previously shown that water, pH and salt concentration impact the glass transition temperature of hydrated PECs. The effect of salt on relaxation times in PECs has been well documented. However, knowledge is lacking on the effect of water and pH on relaxation times in hydrated PECs. This is accomplished by performing time-temperature, time-temperature-water and time-temperature-water-pH superpositions.
In the third study, we discuss the impact of salt concentration on diffusion coefficient of PE in PECs using fluorescent recovery after photobleaching (FRAP) in PECs composed of poly(styrene sulphonate) (PSS) and poly(vinylbenzyl trimethylammonium chloride) (PVBTMA). We first map out the phase diagram at room temperature using optical microscopy, UV-Vis spectroscopy and conductivity measurements. Next, we study the diffusion of counterions using a home-built setup. Finally, we study the lateral diffusion of fluorescently tagged PVBTMA using confocal microscopy.
In the fourth study we discuss the impact of solvent on the lateral diffusion of PVBTMA in PSS-PVBTMA PEMs. First, we study the effect of varying weight % of ethanol and urea on the growth of PEMs. Next, we discuss the effect of varying weight % of ethanol and urea on the strength of interaction between the PEs using Isothermal calorimetry (ITC)
Sheep as a Potential Tool for In-Season Cotton Weed Management
No full textIncreased reliance on herbicides in crop production has led to many weed species becoming
resistant to multiple herbicide modes of action. Sheep herbivory may be a viable alternative
weed control method, as sheep have the potential to preferentially graze weeds and be averse to
eating the cotton plant due to the presence and concentration of gossypol. Common weeds such
as Palmer amaranth and field bindweed are major competitors with cotton plants in western
Texas but are also palatable to sheep. Field research on the integration of sheep into cotton
systems was performed at the Texas AgriLife extension and research center in San Angelo,
Texas during the 2022 and 2023 seasons. Treatments included three different cotton growth
stages to initiate grazing (4-leaf, 8-leaf, and mid-bloom) and three different levels of grazing
intensity based on weed removal (approximately 70%, 90%, and 100%) with presumably greater
cotton damage with increasing intensity. Treatment effects were quantified through monitoring
sheep grazing activity, assessments of weed biomass removal, cotton damage, and cotton yield.
During the 4-leaf, 8-leaf, and mid-bloom initiation for both years, sheep spent 87%, 86%, and
93% of feeding time, respectively, grazing on weeds rather than cotton. Final cotton biomass was
not influenced by year, intensity, or timing of treatments. Final weed biomass was affected by
year (P > 0.069) and timing (P > 0.036). The year 2022 had less final weed biomass than 2023
and grazing initiated at the 4-leaf stage resulted in greater weed biomass at the end of the
growing season when compared to grazing initiated at the 8-leaf stage. This trial emphasizes the
challenge of extrapolating small-scale findings to field conditions, where sheep grazing may
occur at different times. While small-plot research is valuable, its limitations highlight the need
for field-scale observations. Integrating sheep grazing into production systems shows promise for
farmers seeking reduced herbicide/organic management, but further refinement and consideration of economic impacts are necessary. Future research may assess grazing preferences relative to sheep age and breed to provide greater insight into integrated crop-livestock management practices
Advancing Iron Catalyzed Three-Component Cross-Coupling Reactions
No full textTransition metal���catalyzed cross-coupling reactions are some of the most widely used methods in chemical synthesis. Notable advantages of iron as a potentially cheaper, more abundant, and a less toxic transition metal catalyst have drawn the interest of our lab, in particular to explore the mechanism of action in three-component radical cross-couplings. In the first project, we explored the difunctionalization of unactivated olefins with alkyl halides and Grignard reagents. The reaction tolerates a wide range of sp^2 hybridized nucleophiles, alkyl halides, and unactivated olefins bearing a diverse range of functional groups.
Our second work highlights iron���s practical application in more elaborate multicomponent cross-couplings including formation and trapping of ��-boryl radicals and allyl alkyl halides for practical synthesis of cyclic fluorous compounds. Incorporating fluorine into drug scaffolds remains of utmost importance in medicinal chemistry since it generally increases lipophilicity, stability, and overall lifetime, and ~20% of drugs on the market contain at least one C-F bond. In that vein, pinacol boronate esters and boronic acids are excellent building blocks due to their reaction efficiency, low cost, and ability to be transformed into many other desired functional groups.
The final research focus is on using a mechanistic-driven approach towards designing new chiral organoiron catalytic species capable of controlling the C-C bond formation with diverse C-centered radicals. To date, there are only three examples of enantioselective iron-catalyzed cross-coupling reactions, and all are limited to the union of only two components. We reported a practical and simple protocol that uses commercially available and inexpensive iron salts in combination with chiral bisphosphine ligands to enable the regio- and enantioselective (up to 91:9) multicomponent cross-coupling of vinyl boronates, (fluoro)alkyl halides, and Grignard reagents. Preliminary mechanistic studies are consistent with rapid formation of ��-boryl radical followed by reversible radical addition to mono-aryl bisphosphine-Fe(II) and subsequent enantioselective inner-sphere reductive elimination. Overall, my research is expected to expand the field of asymmetric iron cross-couplings and have broad implications towards the synthesis of bioactive compounds via the use of alkenes to translocate alkyl radicals, modify their steric and electronic properties, and induce stereocontrol
Probabilistic Clustering Methods for Complex Data and Related Topics
No full textProbabilistic clustering involves identifying mutually exclusive subsets of data in a probabilistic manner by characterizing the probability distribution on the space of partitions, enabling uncertainty quantification. Random partition models serve as prior probability models for probabilistic clustering, which are closely connected to Bayesian nonparametric models involving discrete random structures. This dissertation aims to contribute to the existing literature on probabilistic clustering and related topics by introducing new key concepts and novel methodologies.
First, we offer new perspectives on exchangeable random partition models based on the concept of balancedness and risk equilibrium/penalization priors with respect to a loss function. The balancedness property provides a better understanding of what probabilistic clustering model works better and what does not for different applications. The notion of risk equilibrium and penalization priors, which are not limited to clustering problems, provide the development of objective and regularized prior model probabilities in light of the effect of loss functions.
Next, we suggest Bayesian hierarchical models with non-exchangeable random partition models associated with a graph that represents complex dependence structures. We start by introducing the low-rank horseshoe prior based on random spanning trees, which simultaneously detects structured sparsity and smoothness of high-dimensional parameters associated with the graph. Also, we introduce the graph product partition model where a probability of partition is defined as a product of cohesion functions capturing internal and external connectivity. This formulation does not rely on graph generative assumptions and can be used as a robust alternative for the probabilistic clustering of graphs with node attributes.
Finally, we propose the logistic-beta process, whose logistic transformation yields a stochastic process with common beta marginals that have significant computational benefits. It has great potential in applications to dependent Bayesian nonparametric models involving beta distributions, related to probabilistic clustering of non-exchangeable data. We illustrate with a Bayesian density regression problem based on the newly proposed logistic-beta dependent Dirichlet process
Bifacial Passive Radiative Cooling of Silicon Solar Cells Using PDMS for Increased Efficiency
No full textWith the rising levels of pollution and environmental degradation due to fossil fuel energy, there is an urgent need for clean and renewable energy. To compete with the vastly greater generation capacity of fossil fuels, the sources of renewable energy need to be more efficient at lower cost. There has been an increasing amount of work in improving the efficiency of solar cells by using several different materials and methods to generate power more effectively. However, all standard solar cells heat up in the course of operation under the sunlight. This heat absorption comes with significant reductions in the efficiency, reliability, and overall lifespan of the solar cells. In this thesis research, we present a relatively inexpensive, energy efficient and simple way to increase the efficiency of a solar cell by using a high emissivity material showing the effectiveness of radiative cooling on the simplest monocrystalline silicon solar cell. We found that coating a high emissivity material, such as PDMS, on both top and bottom of a solar cell allows for a large amount of thermal radiation to be removed to outer space, which acts as the ultimate heat sink. With in-field tests, we found an average temperature reduction of 6��C of a 1.06% performance increase. Considering the limited power conversion efficiency of a typical solar cell, we anticipate that energy efficient radiative coolers will be ubiquitous in the upcoming generations of solar cells
Improving the Ability of Activity Recognition Systems to Detect Activities of Daily Living Performed In-the-Wild
No full textFailing to keep track of the performance of activities of daily living (ADLs) can lead to adverse health outcomes for people with health concerns. However, current recommended practices for keeping track are tedious and burdensome, making it easy for people to forget or stop managing their health. Using activity recognition systems to automatically detect and record ADL performance would address this issue, but most works in activity recognition focus on controlled or semi-naturalistic data in contrast to real world, in-the-wild data. As such, real world ADL recognition remains an open problem. Specifically, real world ADL recognition requires tackling several fundamental challenges for machine learning systems, and it is unclear if existing approaches would be robust to these challenges. We expect that semi-naturalistic data does not capture the diversity of all of the everyday activities such a system would encounter and that robust performance requires using in-the-wild data.
In this work, we focus on quantifying the challenges associated with in-the-wild settings and investigating the design of in-the-wild ADL recognition systems. To achieve these goals, we conduct a series of analyses and machine learning experiments on two ADL datasets, one semi-naturalistic and one in-the-wild. First, we measure the class imbalance, interpersonal variability, and pairwise class overlap to motivate the difficulty of recognizing in-the-wild data. Second, we demonstrate the importance of training on negative samples, showing that training on NULL data results in more robust models than using unknown class rejection. Third, we investigate the design of in-the-wild ADL recognition systems, exploring both classical and deep learning methods as well as models with varying levels of context of the user���s hands. In doing so, we develop a recognition system that can recognize several ADLs with high event-based recall and precision with only the context of the dominant hand. These efforts represent a thorough investigation of a challenging open problem in human activity recognition. The results and insights serve as a meaningful step forward toward making robust in-the-wild ADL recognition a reality in order to make it easier for people to manage their health
Innovative Fabrication of Asymmetric Mixed-Matrix Membranes for Gas Separation
No full textDespite extensive research on mixed-matrix membranes (MMMs), their path to commercial deployment has been obstructed by a range of scientific and engineering barriers. This study introduces a novel, one-step method for fabricating MMMs, promising to advance their commercialization by addressing many of these existing challenges. This method, known as phase-inversion in sync with metal-organic framework (MOF) formation (PIMOF), allows for the swift creation of high-performance, asymmetric MMMs on a scalable basis. It involves a polymer film undergoing phase inversion while ZIF-8 nanoparticles are simultaneously generated in-situ within the polymer matrix. The produced MMMs exhibit exceptional propylene/propane separation efficiency, largely owing to the high effective ZIF-8 loading and the augmented molecular sieving capability of the uniquely small sub-5 nm ZIF-8 filler nanoparticles, facilitated by restricted linker swing motion. Subsequently, our research delves into the impact of modulating ligands, sodium formate and 1,4-butanediol, on the development of ZIF-8 fillers within the polymer matrix, aiming to elevate the separation capabilities of asymmetric MMMs crafted via the PIMOF technique. Our computational analysis supplements this empirical research, examining how the size of ZIF-8 nanoparticles influences the specific surface interaction energy and the apertures of ZIF-8. The findings suggest that smaller ZIF-8 nanoparticles foster stronger interactions with the polymer, affecting the nanoparticle's aperture size. These advancements mark a significant leap forward, surpassing the performance of all previously documented propylene-selective MMMs. Lastly, a hybrid strategy that involves the mixing of metals and/or linkers has been explored to precisely adjust the porosity and surface characteristics of the ZIF framework, thereby broadening their utility in the separation of various critical gas mixtures. The technique of linker-doping, as detailed in this report, represents a viable approach to diversify the range of imidazolate linkers, enabling the creation of mixed-linker hybrid ZIFs. Specifically, the doping of 2-ethylimidazole (eIm) into the frameworks of ZIF-8 and ZIF-67 has been examined, demonstrating the potential of this innovative method to improve gas transport across the membranes. This approach not only illustrates the adaptability of ZIF materials to meet diverse separation needs but also highlights the potential for tailored membrane design to enhance separation performance
A Study of Texture Characterization of Fiducial Markers for Visual Navigation
No full textA full six degree-of-freedom state estimation is an important problem in robotics, augmented reality, and autonomous navigation. Obtaining such information using visual features has been challenging since such a task always needs information-rich images. Getting access to a visually rich environment is not often possible. In such scenarios, artificial visual references like fiducial markers are used. This thesis conducts a systematic study of the texture of such fiducial markers. The study provides insight into the fiducial markers��� visual characteristics and design patterns. A general implementation of the ArUco marker detection and estimation system is created to understand the fiducial marker design process fully. The thesis reports results and lessons from both the set of tasks. Based on the study of other markers, a new fiducial marker-based on the Spidron pattern is proposed. A detection and pose estimation system is developed for this Spidron based marker. The system is tested in an experimental setup simulating refraction, motion blur due to rotation and jitters, and transformation due to scaling
Sequence Stratigraphy and Chemostratigraphy of the Middle Cretaceous Woodbine and Eagle Ford Groups on the Southeastern Margin of the East Texas Basin and East Texas Submarine Plateau
No full textWithin the East Texas Basin (ETB), the Woodbine and Eagle Ford groups are important hydrocarbon reservoirs. However, little work was done within the southeast margin of the ETB, and adjacent East Texas Submarine Plateau (ETSP) region, to properly differentiate these units. In fact, the entire succession between the Buda Formation and Austin Groups often is referred to as the ���Eaglebine���. When defined, sandstone beds, as well as underlying mudstone beds, within it, are simply lithologically assigned to the Woodbine Group. This study takes a surface-based, or sequence stratigraphic, approach using: 1) a grid of well log cross sections, 2) x-ray fluorescence (XRF) data from cuttings and cores, and 3) previously published seismic data, to chronostratigraphically define and differentiate the Woodbine and Eagle Ford Groups across the study area.
Across much of the western portions of the study area, in Grimes and Madison Counties, a regional unconformity at the base of the Lower Eagle Ford Formation (K630sb) separates the Woodbine Group below from the Eagle Ford Group above. In this area, TOC- and Ca-rich, high-resistivity mudstone of the Lower Eagle Ford Formation unconformably overlie Al-rich, moderate resistivity mudstones of the Pepper Shale, which represent the distal downdip (basinal) equivalents of the Woodbine Group Freestone Delta.
In the western portions of the study area, in Houston and Walker Counties, a regional unconformity at the base of the Upper Eagle Ford Formation (K650sb), truncates the Lower Eagle Ford Formation near the interpreted K630 depositional shelf break. In this area, Al-rich, low-resistivity mudstone, and overlying sandstone, of the Upper Eagle Ford Formation Harris Delta, unconformably overlie Al-rich, low-resistivity mudstones, and overlying sandstones, of the Woodbine Group Freestone Delta. Fortunately, a low-resistivity marker, interpreted as regional flooding surface (K650mfs), near the base of the Upper Eagle Ford Formation can be defined, mapped, and used to differentiate strata of the Upper Eagle Ford Formation Harris Delta from the underlying Woodbine Group Freestone Delta in this area.
Finally, in the southeastern portions of the study area in Polk County, a regional unconformity at the base of the Austin Group (K720sb) truncates most, or all, of the Eagle Ford Group, and the Austin Group, or the lowermost portion of the Upper Eagle Ford Formation unconformably overlie the Woodbine Group Freestone Delta. A thick succession of the Eagle Ford Group was deposited downdip of the interpreted K630sb depositional shelf break in this region