Master of Science (MS), Bowling Green State University, 2022, Physics

While Silicon photovoltaic (PV) cells have been the workhorse of solar power generation, many different PV technologies will need to be employed to achieve a greener energy future. One promising emerging technology is Perovskite solar cells, which have demonstrated >25% power conversion efficiency (PCE) with only a decade of research. It has been found that this material also has desirable properties for another emerging technology: All-Back-Contact (ABC) thin-film PV devices. This work investigates Perovskite absorbers in two different ABC designs: Quasi-Interdigitated Back Contact (QIBC) and Metasurface Lattice Back Contact (MSLBC) consisting of a periodic array of photo-absorbing cylinders. Due to the similarity to MSLBC designs, a Metasurface Perfect Absorber (MSPA) was also investigated. The optoelectronic performance of these devices was simulated using COMSOL Multiphysics®, which uses the finite element method to solve sets partial differential equations. It was found that a textured front surface for the QIBC device provides superior optical absorption over a flat front surface at higher wavelengths due to light trapping and optical resonance. Despite the inferior optical performance, the flat front surface QIBC design was shown to achieve a PCE >23%. The MSPA design was found to have high absorption for a wide range of incident angles due to Mie resonance within the periodic cylindrical structures. Strong absorption was also observed in the MSLBC device due again to Mie resonance. The PCE of this device was found to be superior under bottom illumination with room for improvement. Optimization of this design was performed by targeting the cylinder diameter and cylinder height. By comparing the optical and electronic performance for a wide range of these parameters, optimal values were found. The bifacial performance of this device was then examined and it showed little improvement over single-side illumination.

Doctor of Philosophy (Ph.D.), University of Dayton, 2019, Electrical and Computer Engineering

Magnetron oscillator based marine radar technology is mature, affordable, reliable, and very effective for maritime safety applications. Commercial systems such as the Furuno DRS25A may be procured at a modest cost as compared to fully coherent solid-state systems costing many tens of thousands of dollars. Magnetron oscillators inherently generate random phase signals. Phase instability on a pulse-to-pulse basis impedes this class of marine radar from success in applications requiring coherency such as moving target indication (MTI) or in generating target imagery. This limitation may be overcome by incorporating radio frequency (RF) sampling technology to augment the current capability of available systems. RF sampling in order to extract a reference signal on transmit and a target echo on receive permits fully coherent processing. Marine radars traditionally operate non-coherently, and as such, offer limited surveillance in clutter rich environments. In this article, we report on a non-coherent marine radar that has been modified to produce a pseudo-coherent (coherent-on-receive) sensor system. This is crucial to MTI or target image formation. In laboratory experiments, we employed a magnetron oscillator based system to generate an inverse synthetic aperture radar (ISAR) image. Using four different algorithms: filtered back-projection (FBP), time domain and frequency domain back-projection (TDBP and FDBP), and the Algebraic reconstruction technique (ART), images have been formed and results compared.

Doctor of Philosophy, University of Toledo, 2021, Physics

Polycrystalline Cadmium Telluride has been developed to be one of the most commercially successful materials for photovoltaic module production with power conversion efficiencies over 21% for research cells to over 18% for module efficiencies. However, little is known about these record devices architecture or the processing methods. Following conventional understanding of a CdTe solar cell operation, researchers have put extensive efforts over the years to improve the CdTe device performance through improved material quality and diode quality. While this have gained some benefit, performance limiting factors to these devices remains unchanged. Deviating from conventional concepts, better understanding of the device physics is needed in order to further improve these devices. This dissertation focusses on identifying these loss mechanisms and setting guidelines to fabricating high efficiency CdTe devices through both experimental and numerical simulation. Experimental work discusses the details to construction and characterization of a CdTe deposition system and employing the new understanding of improving the CdTe device to achieve high performing CdTe devices. Here the traditional CdS window layer is replaced by a wide bandgap MgxZn1-xO to increase the photocurrent generation with better band alignment. With optimum deposition and processing conditions, work demonstrates a device with power conversion efficiency >16%. With a good front contact, performance of the device can be limited by the poor back contact. Expanding the understanding to front contact band alignment, characteristics of a back buffer layer suitable for CdTe back contact is also explored. Through 1D numerical simulation of the conduction and valence band offset, doping levels of the CdTe and back buffer layer material, this dissertation work sets the guideline to achieving CdTe device performance up to 25%.

Master of Science, The Ohio State University, 2020, Electrical and Computer Engineering

As an emerging area of research, near-field microwave imaging has elicited considerable interest. The remarkable properties of microwaves and the convenience of portable imaging systems make this technology attractive in various application areas. This thesis conducts the comparison between coherent and incoherent imaging algorithms: back projection algorithm, interferometric imaging algorithm and self-correlation back projection algorithm for the near field circumstance. Near-field interferometric imaging is a relatively new incoherent approach that can use virtually any microwave source of opportunity rather than a conventional radar transmitter. A thorough comparison with the coherent back projection approaches has not been previously conducted. Notwithstanding merits of microwave imaging, the microwave frequency band is vulnerable to the environment. In addition, the received signal is mixed with the inevitable Direct Path Interference (DPI) from transmitter during the detection process. The DPI signal can be orders of magnitude stronger than the echo signals from the targets of interest, causing an unsatisfactory construction with the desired targets lost in the interference sidelobes. This thesis investigates ways to mitigate this effect via different signal processing methods. Matlab simulated scenarios are used to illustrate the problem and solutions. Further, this thesis investigates the lateral and range resolution of the back projection, interferometric imaging and self-correlation back projection method. The results are validated by simulation of imaging scenarios through CST and Matlab.

Master of Science, The Ohio State University, 2019, Mechanical Engineering

In this study, a generalized methodology is designed and implemented to characterize dynamic behavior of a certain class of gear test machines, namely back-to-back setups. This methodology relies on two independent measurement systems. One system focuses on measurement of tooth root strains to be directly correlated to dynamic gear mesh forces transmitted by the gear mesh. A compliantly connected slip ring system is incorporated with this measurement system for noise-free transmission of strain signals to a stationary DAQ. A data analysis procedure is developed to process the strain data to determine dynamic stress factors within a range of speed and torque under both steady-state and transient operating conditions with appropriate filtering and statistical analysis procedures applied. The other measurement system uses an array of accelerometers mounted at stationary locations along the test gear pair bearing caps. The methodology is adapted to two separate back-to-back gear test machines, one machine being an ISO standard FZG machine, and the other being a newly developed GL-100 machine. The measurements from both machines are compared at the end to determine that the difference in their dynamic behavior is minimal.

Master of Science, The Ohio State University, 2019, Mechanical Engineering

In this study, a new state-of-the-art gear contact durability test machine is developed to eliminate some key limitations, issues and inconveniences associated with conventional FZG machines. Based on lessons learned from prior in-house studies using conventional FZG machines, a number of requirements are specified. A new four-square concept is designed and fabricated to gain various advantages with respect to foot-print, temperature regulation, and long-lasting auxiliary components. The mechanical layout and out-of-the-loop torque application methodology are described along with the heat management and lubrication systems. Various new safety provisions are highlighted, in addition to a streamlined interim gear inspection procedure during long-cycle contact fatigue tests performed on these new machines. For demonstration purposes, the proposed machines are used to conduct an example contact fatigue test program to evaluate a stress-life curve of ground spur gears made of a typical automotive gear steel. The test procedures, test conditions and the test specimens are described. Various results containing digital images of the gear teeth, surface wear and roughness progression over the fatigue life of the gears are detailed. A detailed statistical analysis is presented to define a stress-life curve and confidence intervals. This example fatigue study confirms the suitability of the new machines to perform high-fidelity gear contact fatigue experiments.

Master of Science, The Ohio State University, 2012, Mechanical Engineering

As personalized biologically-assisted models of the spine have evolved, the calibration of raw electromyographic (EMG) signals has become extremely important. The traditional method of normalizing myoelectric signals relative to subjective maximum voluntary contractions (MVCs) is believed to introduce error since true MVCs are commonly underrepresented. Additionally, this method is useless for assessing the biomechanics of symptomatic low back pain (LBP) patients since they are generally unable or unwilling to provide MVCs. Therefore, the objective of this study was to develop and validate a biologically-assisted model calibration technique that does not require MVCs for EMG normalization. The newly developed technique eliminates the need to collect MVCs by combining gain (maximal strength per unit area) and MVC into a single muscle property (gain ratio) that can be calculated during model calibration. An experiment was conducted with ten subjects (5 male, 5 female) to compare the performance and muscle property predictions from this new calibration technique to the traditional MVC-based technique. Models calibrated by each technique were then used to evaluate independent test sets of multi-planar torso exertions and asymmetrical lifts, and resultant validity measures and spinal loads were compared. No statistically significant differences in spinal loads were found between calibration techniques, though differences were observed in muscle properties for the right latissimus dorsi. Model validity measures indicated that the new calibration technique performed at least as well as MVC-based calibrations. This new calibration technique will facilitate personalized muscle property and spinal load predictions for individuals who are not willing or are not able to provide reliable MVCs. In particular, this technique may permit future efforts to evaluate symptomatic low back pain (LBP) patients, which could provide significant insight into the underlying nature of the LBP (open full item for complete abstract)

Master of Science, The Ohio State University, 2010, Mathematics

The problem of collisions of shockwaves in gravity is well known and has been studied extensively in the literature. Recently, the interest in this area has been revived trough the anti-de-Sitter space/Conformal Field Theory correspondence (AdS/CFT) with the difference that in this case the background geometry is Anti de Sitter in five dimensions. In a recent project that we have completed in the context of AdS/CFT, we have gained insight in the problem of shockwaves and our goal in this work is to apply the technique we have developed in order to take some farther steps in the direction of shockwaves collisions in ordinary gravity. In the current project, each of the shockwaves correspond to a point-like Stress-Energy tensor that moves with the speed of light while the collision is asymmetric and involves an impact parameter (b). Our method is to expand the metric (gμν) in the background of flat space-time in the presence of the two shockwaves and compute corrections that satisfy causal boundary conditions taking into account back-to-back reactions of the Stress-Energy tensor of the two point-like particles. Therefore, using Einstein's equations we predict the future of space-time using the fact that we know the past geometry. The expansion we construct is valid for large transverse (to the initial direction of motion of the initial point-like particles) distances (r) and large proper times (τ) compared to the energy carried by the shockwaves. Our solution respects causality as expected but this casual dependence takes place in an intuitive way. In particular, gμν at any given point r⃗ on the transverse plane at fixed τ evolves according from whether the propagation from the center of each of the shockwaves or from both shockwaves has enough proper time (τ) to reach the point under consideration or not. Simultaneously around the center of each shockwave, the future metric develops a δ-function profile with radius τ; therefore this profile expands outwards from the (open full item for complete abstract)

Doctor of Philosophy, The Ohio State University, 2004, Agricultural, Environmental and Development Economics

In the United States, rabies strains that infect raccoons have been responsible for the largest increase animal rabies in the past 3 decades. This work includes three articles that analyze: 1) the cost of 8 distributions of oral rabies vaccine (ORV) with strains known to infect raccoons in Ohio between 1997 and 2000, 2) an agent-based simulation of uninterrupted raccoon rabies epidemic in a hypothetical area, and 3) the costs and benefits of different ORV distribution strategies. Article 1 documents the estimated cost of implementing an ORV program to provide a more efficient use of resources to control and limit the spread of rabies. Accurately measured distribution costs can be used to perform an economic cost-benefit analysis for alternative ORV programs. The existing ORV procedure consists of distributing fishmeal bait containing ORV through various means. The cost of personnel, vehicles, and helicopter and aircraft use and other associated expenses were obtained from field records and interviews with personnel and agencies involved in the ORV program. Article 2 examines the major characteristics and behavior of raccoon agents and their relation to their environment. Under different parameter values, the models are simulated and results of a hypothetical raccoon rabies event is obtained in terms of the rate of disease movement, shape of the epidemic front and intensity of new infections. The results indicate that model results are sensitive to certain parameters (e.g., aggressiveness of the epidemic regime, or nutrient regeneration capability of spatial units). Results on the shape of epidemic front proved to be invariant to different selection of model parameters. In article 3, different ORV distribution strategies were devised to assess the effectiveness of ORV distribution strategies under different assumptions and their potential costs. Based on raccoon rabies literature, incidences of new infections were mapped to economic costs. These costs were used in co (open full item for complete abstract)

Doctor of Education (EdD), Wright State University, 2024, Leadership Studies

This dissertation focused on a novel improvement to the current method for adapting assessments into American Sign Language (ASL). Bilingual Deaf adults participated in back translations across the United States, and those back-translation decisions were assessed by human experts to judge similarity in meaning. Translations were compared to original text samples using two types of Latent Semantic Analysis (LSA) models to compute semantic textual similarity (STS) scores, and to calculate weighted Youden Index (WYI) Scores. These scores were used to determine the ideal cutoff to be used when making judgments and compared to human expert decisions. The results revealed that WYI scores calculated using the Bidirectional Encoder Representations from Transformers (BERT) model performed best and effectively predicted expert decisions for 25% of items, thus substantially reducing the need for human review for many items. These results suggest that while there is great promise for using these methods to reduce cognitive load for back-translation tasks, there is still a crucial need for human attention in such tasks. This research points to the potential of machine learning for streamlining the creation of ASL assessments and increasing accessibility for the Deaf community. However, it also underscores the essential role of human experts in ensuring accuracy and cultural sensitivity. While future advancements in machine learning may one day replicate similar human capabilities, a combination of technology and skilled professionals remains crucial for bridging these communication gaps and providing equitable access to services for Deaf individuals.

Master of Science, The Ohio State University, 2005, Graduate School

Master of Science (MS), Bowling Green State University, 2024, Physics

Cd(Se,Te) has emerged as a leading choice for commercial thin-film PV devices, owing to their lower cost of production, high energy yields, and low degradation rates compared to silicon technology. Despite significant advancements, Cd(Se,Te) cells suffer from recombination losses, reducing the open-circuit voltage (Voc). This thesis aims to identify, distinguish, and quantify recombination losses and their locations within Cd(Se,Te) solar cells via temperature and light intensity-dependent current-voltage (JVTi) analysis. Cd(Se,Te) solar cells were modeled using COMSOL Multiphysics, simulating parameters such as temperature (T ), light intensity (i), front surface recombination velocity (Sf), back surface recombination velocity (Sb), bulk lifetime (τ ), conduction and valence band offset (CBO and VBO at heterojunctions), and back contact Schottky barrier height (Φbp). Additionally, graded and uniform selenium devices were studied, and ZnTe:Cu was investigated as a back contact interface. In this work, recombination activation energies, Ea, from JVTi studies were shown to quantify the front interface conduction band offset losses when the interface band gap is smaller than the bulk band gap and when front interface recombination dominates. If the Ea equals the bulk band gap, then Voc losses may occur at the front interface or within the bulk. When the front surface recombination and bulk lifetime are moderately low, a transition from front surface (low Ea) to bulk (higher Ea) mechanisms can be observed with increasing light intensity, i. Back surface recombination has negligible effects on Voc for the device parameters specified herein. Comparison of Cd(Se,Te) JVTi data provided by NREL to simulations in this work indicates that front surface recombination dominates Voc losses for Sf = 103 cm/s and CBO = -0.2 eV for that particular device. Adjusting the band alignment to CBO = 0 eV and reducing Sf would significantly increase Voc.

Doctor of Philosophy (Ph.D.), Bowling Green State University, 2024, Statistics

Deep learning, neural network, has been penetrating into almost every corner of data analyses. With advantages on computing power and speed, adding more layers in a neural network analysis becomes a common practice to improve the prediction accuracy. However, over depleting information in the training dataset may consequently carry data noises into the learning process of neural network and result in over-fitting errors. Neural Network has been used to predict the future time series data. It had been claimed by several authors that the Neural Network (Recurrent Neural Network) can predict the time series data, although time series models have also been used to predict the future time series data. This dissertation is thus motivated to investigate the prediction performances of neural networks versus the conventional inference method of time series analysis. After introducing basic concepts and theoretical background of neural network and time series prediction in Chapter 1, Chapter 2 analyzes fundamental structure of time series, along with estimation, hypothesis testing, and prediction methods. Chapter 3 discusses details of computing algorithms and procedures in neural network with theoretical adjustment for time series prediction. In conjunction with terminologies and methodologies in the previous chapters, Chapter 4 directly compares the prediction results of neural networks and conventional time series for the squared error function. In terms of methodology assessment, the evaluation criterion plays a critical role. The performance of the existing neural network models for time series predictions has been observed. It has been experimentally observed that the time series predictions by time series models are better compared to the neural network models both computationally and theoretically. The conditions for the better performances of the Time Series Models over the Neural Network Models have been discovered. Theorems have also been pro (open full item for complete abstract)

Doctor of Nursing Practice, Mount St. Joseph University , 2024, Department of Nursing

Heart failure is the leading cause of readmissions among chronic illnesses in the United States (Oh et al., 2021). Patients with chronic disease need correct knowledge and understanding about maintaining, monitoring, and managing their illness in their daily lives. However, patients with low health reading ability have a more challenging time with comprehension (Oh et al., 2021). It is imperative for acute care nursing staff to gain the ability to properly educate heart failure patients on the disease process to increase the patient's quality of life while decreasing their chance of hospital readmission. This project aimed to increase nursing knowledge related to heart failure and disease-specific education using smart-phrase technology to reduce heart failure 30-day readmission rates. After initial stakeholder meetings, staff education and training, and project implementation, the project did show a decrease in hospital admissions within the 12-week utilization period. The readmission data is under review to ensure the smart-phrase is still effective in current practice. The project achieved its goals, including improving nurse competency in heart failure and simplifying patient education.

PhD, University of Cincinnati, 2023, Medicine: Molecular, Cellular and Biochemical Pharmacology

Inflammatory low back pain (LBP) is an extremely common and debilitating condition that is often treated with corticosteroids, which activate the glucocorticoid receptor (GR) in sensory neurons of the dorsal root ganglion (DRG) to reduce inflammation. However, many clinically used corticosteroids administered for LBP can also activate the mineralocorticoid receptor (MR) with similar or even greater potency. In rodent inflammatory pain models, MR antagonists reduce pain behavior, neuronal excitability, and inflammatory mediators. In this work (Chapter 2), we have compared human and mouse DRG expression of the GR and MR transcripts, finding that these receptors are almost ubiquitously expressed and colocalized in sensory neurons. The GR is predominant in nociceptors, while the MR is predominant in Aβ mechanoreceptors across species. This finding bodes well for translation of rodent corticosteroid research to clinical pain treatments, and suggests important roles for GR & MR in specific neuronal subpopulations. We have also shown (Chapter 3) that MR antagonism reduces mechanical hypersensitivity in rodents in an inflammatory LBP model, but MR KO in sensory neurons does not. While sensory neuronal MR KO does not change pain behaviors, it does change neuronal excitability and GR expression. These findings support the use of MR antagonists clinically, but show that sensory neuronal MR KO is not having the same effect. We learned through the MR KO model that MR has opposing roles in A- and C- neuronal excitability, and directly or indirectly regulates GR expression before and during inflammatory LBP. This research is the first to use mice as a model for corticosteroid receptor regulation in pain, shows translational possibility, and furthers mechanistic insights into corticosteroid receptor action at the DRG using genetic manipulation techniques.

Doctor of Philosophy, The Ohio State University, 2023, Biomedical Engineering

Low back pain (LBP) is a leading cause of disability worldwide, affecting approximately 80% of adults at some point in their lifetime, and has large associated socioeconomic burdens. Intervertebral disc (IVD) degeneration is a major cause of LBP, accounting for approximately 40% of all LBP cases, yet current clinical therapies do not target the underlying pathology that IVD cells experience in degeneration. While cell therapies and viral gene delivery have been of interest in IVD regeneration, their clinical application is limited due to caveats such as cell survivability, mutagenesis, and unwarranted immune response. Thus, a method of non-viral delivery of prominent developmental transcription factors into diseased patient cells may serve as a novel, minimally invasive, and non-addictive therapy for IVD degeneration utilizing the patient's own cells. The first chapter of this dissertation covers the background on the clinical relevance of LBP and IVD in its healthy and diseased states along with detailed background on current therapies for IVD degeneration. This then leads into the advantages of cellular reprogramming via non-viral delivery and developmental transcription factors of interest (FOXF1, Brachyury, Mohawk, Scleraxis) along with a brief review of animal models relevant to IVD degeneration. Chapter 2 explores the non-viral delivery of Brachyury, a transcription factor critical to the development of the nucleus pulposus (NP) region of the IVD, in diseased human NP within a physiologically relevant 3D culture model. Similarly, Chapter 3 assesses the non-viral delivery of FOXF1, a key phenotypic marker of healthy NP cells, into diseased NP cells via electroporation and engineered extracellular vesicles (EVs) in both human cells in vitro and mice IVDs in vivo. Chapter 4 explores the outer annulus fibrosus (AF) region of the IVD via non-viral transfection of Mohawk or Scleraxis into diseased human AF cells. These chapters are then summarized in the final concl (open full item for complete abstract)

Master of Science, Miami University, 2023, Computer Science and Software Engineering

It is a well-established fact that the performance of Machine Translation (MT) techniques largely depends on the quantity and quality of data available. The lack of a large well-curated dataset is especially a challenge for low-resource languages. The Myaamia language, also known as the Miami-Illinois language, is an endangered Native American language, and there are active efforts being made toward its revitalization. As a part of the revitalization process, the recorded texts are currently being manually translated, which might take up to a decade to translate at the current rate, according to some expert assessments. To speed up the translation process, we developed Myaamia Translator, a Neural Machine Translation (NMT) based machine translation approach, which leverages the state-of-the-art transformer architecture to translate text from Myaamia to English. The contributions of this work are two-fold: first, we use a combination of rule-based augmentation and back-translation augmentation to address the data limitation; and second, we train the model using the large dataset to test its effectiveness in translating a religious Myaamia textbook to English.

Master of Science, The Ohio State University, 2023, Industrial and Systems Engineering

Introduction: Musculoskeletal disorders in the workforce are highly prevalent, especially in material handling operations. In addition to completing physically demanding work that is required in this domain, workers must also manage concurrent mental demands present in their tasks. Few studies have examined the effect of concurrent mental demands in occupationally-relevant tasks. This study attempted to fill this void by quantifying the effects of varying degrees of cognitive loads and task precision demands on a material handling task by examining these effects on the kinematics and muscle activity of the trunk and shoulders. Methods: Twelve subjects lifted and placed a 5 kg box on a rack at one of three destination heights (low, middle, high) while under a simultaneous cognitive load (no load, simple load, complex load) and/or precision constraint (low precision, high precision). Cognitive load consisted of time-based arithmetic questions where participants were tasked with determining the amount of time remaining from a given time to a target time (e.g., Get to 4:00 PM from 3:15 for simple load or get to 4:10 PM from 3:27 PM for complex load). The primary dependent measures were the angular velocities of the trunk and shoulders as well as muscle activity in the erector spinae, rectus abdominus, external oblique, latissimus dorsi, and anterior deltoid muscles. Results: Significant decreases in angular velocities for both higher cognitive load complexities and higher precision conditions were observed. Additionally, lower 90th percentile normalized muscle activity values were observed as complexity and precision increased. Cumulative muscle activity, however, increased with these increases in complexity and precision. Conclusions: This study examined the impact of varying levels of cognitive and precision conditions on muscle activity and kinematics of the trunk and shoulders. Results indicated that increased complexity and precision led to longer lift t (open full item for complete abstract)

Doctor of Philosophy, University of Toledo, 2022, Physics

Solar energy is a powerful and essential source of renewable energy that works on the principle of the photovoltaic (PV) effect. Among all photovoltaic cell technologies, silicon (Si) dominates with a market share of ~ 90% of the PV industry. Though Si is highly efficient, cadmium telluride (CdTe) competes very effectively with crystalline Si (c-Si) for utility-scale PV and offers advantages in terms of energy payback time (EPBT) and low water impacts. Antimony sulfide (Sb2S3) absorber is an exploratory absorber material with favorable properties like earth abundance, a tunable band gap, and a high absorption coefficient. It also has the potential to serve as a top cell absorber for tandem PV technologies. CdTe is an II-VI semiconductor material with a direct band gap of 1.45 eV. Single junction polycrystalline CdTe solar cells have reached a certified photoconversion efficiency (PCE) of 22.1%. According to the Shockley-Queisser limit, for a material having a band gap of 1.45 eV, the theoretically attainable efficiency is 33.1%, suggesting ample room to improve the CdTe device's efficiency. The low PCE in CdTe is primarily due to the device's low open circuit voltage (VOC) and fill factor (FF). CdTe has a deep valance band edge of 5.9 eV below the vacuum level, that creates a barrier at the back interface, limiting hole transport and reduces the device's performance. The suitable implementation of a back buffer layer between the CdTe absorber and the back electrode is vital in improving monofacial and bifacial CdTe device performance. This dissertation addresses the synthesis and characterization of tellurium (Te) and lead telluride (PbTe) nanowires (NWs), copper iodide (CuI) nanoparticles, copper chromium oxide (CuxCryOz), and their applications to fabricate monofacial and bifacial solar cells based on CdTe as well as CdSexTe1-x absorber layers. In addition, the dissertation presents the fabrication and characterization of Sb2S3 solar cells. Tellurium and lead t (open full item for complete abstract)

Master of Science, The Ohio State University, 2022, Industrial and Systems Engineering

While biomechanical models can be insightful to potential mechanisms that can contribute to low back pain, they require significant computing power and various technologies that are not readily available in settings such as clinics. On the other hand, lightweight sensing systems such as inertial measurement unit sensors (IMUs) can be integrated in wearable technologies to easily capture high level spine kinematics. While kinematics can be useful for identifying pain populations or tracking motion metrics for patients, they do not provide a look at the internal structures of the spine during dynamic motions. Several studies have attempted to bridge this disparity. Specifically, muscle coactivity of the trunk muscle groups can greatly impact the overall loading and shear forces on the spine, and several studies have attempted to only use kinematics to predict the trunk muscle forces during static exertions. The goal of this study was to present a methodology that can predict trunk muscle forces as a time series, during dynamic motions, which does not require electromyographic (EMG) signals. This was achieved by first, collecting a new dataset with motions that capture key spine kinematics using an EMG-assisted biomechanical model. Second, to use kinematic-derived variables from this data with deep learning methodologies to predict the trunk muscle forces. 30 healthy subjects performed a series of unloaded, bending, and twisting, during a standardized spine motion assessment while wearing EMG sensors on ten trunk muscles. Several variables were extracted from the time series, including velocities, muscle lengths, height, weight, and torso angles. Using several deep learning architectures, these variables were trained to map these variables to the ten muscle forces produced during the dynamic motions. Various deep learning architectures were investigated, however only three main architectures were reported and pursued. Assessed on an independent test set, the architec (open full item for complete abstract)