ABSTRACTS for CLASS of 2025 SENIOR PROJECTS

as of September 23, 2024

These abstracts describe work that students PLAN to do in their projects; the abstracts for their final papers will generally be very different.

Omar Ali with Prof. Paola Cappellaro (MIT)

Decoherence dynamics in hyperfine-coupled solid-state quantum gyroscopes

Gyroscopes based on solid-state electronic and nuclear spin systems have been proposed as highly sensitive rotation sensors with applications in both fundamental physics and navigation technologies. Such systems leverage the ability to manipulate and measure both the electronic and nuclear spins whose quantum states respond differently to physical rotation, allowing precise measurement of rotation rates by projecting their evolving states into each other. However, when the two spins are strongly coupled, their intrinsic decoherence dynamics—such as relaxation and dephasing—are altered. Previous studies have examined this coupling-mediated decoherence using a ‘spin-fluctuator’ model, where the decoherence of one spin is studied under a fluctuating magnetic field that follows a classical random telegraph noise (RTN). However, under physical rotation, both the amplitude and direction of these magnetic field fluctuations experienced by one spin due to the other evolve dynamically at different timescales, leading to different regimes of coupling-mediated relaxation and dephasing dynamics. This work aims to develop minimalistic models that captures these rotational dynamics and elucidates their dependence on control parameters such as the external magnetic field, ultimately improving the understanding of the fundamental limits of hyperfine-coupled solid-state gyroscopes and guiding techniques and protocols for controlling them optimally.

Sophia Buffone with Dr. Ina Martin and Dr. Andrew Lininger 

Elucidating Degradation Modes in Silicon Architectures

As a response to the increasing need for solar power, new photovoltaic cell architectures have been recently introduced that exhibit record efficiencies. However, the implementation of new materials combinations within these structures raises concerns about their durability and reliability in the field. The silicon heterojunction (SHJ) cell architecture is composed of interfaces unique and essential to its energy production. Unlike conventional designs, the SHJ includes thin layers of intrinsic hydrogenated amorphous silicon (a-Si:H) to reduce the surface recombination rate by passivating the crystalline silicon (c-Si) layer. Its top layer is a transparent conductive oxide (TCO), which supports lateral transport of charge carriers and doubles as an anti-reflective coating (ARC). To study changes in these layers, samples will be aged in accelerated weathering chambers that simulate components of outdoor conditions, specifically a damp-heat and an isolated UV environment. Characterization methods such as Fourier-Transform Infrared Spectroscopy (FTIR), X-ray Photoelectron Spectroscopy (XPS), and Secondary Ion Mass Spectroscopy (ToF-SIMS) will be used to analyze changes in layer composition after aging intervals. Each technique, with unique considerations and insights into structural properties, will be evaluated to maximize precision and efficient data collection. Several cycles of analysis will be conducted to identify degradation patterns in SHJ film stacks and full cell architectures.

Patrick DeLuca with Prof. Robert Brown and Dr. Robert Deissler

Utilizing CAPGLO as an Immunotherapy Device

For my senior project, I will be working on the CAPGLO project alongside Dr. Deissler under the direction of Professor Brown.  The CAPGLO project is an ongoing project aimed at creating a low-cost solution to address a need for an easily accessible and portable cellular capture and detection device.  The project utilizes magnetic beads and fluorescent dye to mark cells of interest.  These cells are concentrated via magnetic force and imaged using fluorescent microscopy.  My goal for my senior project is to successfully implement the CAPGLO device as a tool for cancer t-cell therapy.  Over the next year, we plan on implementing our current methodology on cancer t-cells through a collaborative effort with the medical school, using our current methods to concentrate t-cells with the aim of developing effective immunotherapy.

Collin Dowhan with Prof. Corbin Covault

Muon Telescope

I will be working with Professor Corbin Covault to create and test a muon telescope that utilizes both scintillation detection and Cherenkov radiation in coincidence. Through NIM Logic units and lab-made discriminators, top and bottom scintillation pads are put in coincidence to filter out any events occurring in the upper pad that do not have enough energy to make it to the secondary lower pad. Between the two pads is a large water tank (made from a trashcan), that is lightproofed and closed with top hat electronics including a Photo Multiplier Tube that measures the Cherenkov radiation events. Testing of the entire apparatus to simply count the number of vertical muons that pass through all three detectors and lowering/determining the size of the error bar to apply to the energy resolution of the central water tank detector. 

Anil Driehuys with Prof. John Ruhl

Galactic Foregrounds Investigation

The cosmic microwave background (CMB) is a very faint signal that provides insights to the early universe. In recent history, there have been sky surveys completed to map the variations in the CMB. Since the signal is so faint, there are many obstacles in the way of accurately measuring the CMB which are referred to as the foregrounds. This experiment will explore various methods of removing the galactic foregrounds signal via simulation techniques, allowing for more clear analysis of the CMB.

Rae Dugger with Prof. Peter Hore (Chemistry, Oxford University) and Prof. Lydia Kisley

Field Inversion Asymmetry of Dynamic Radical Pair Systems

Migratory birds use a magnetic inclination compass to sense the geomagnetic field, which is proposed to function utilizing photochemically induced radical pairs. Behavioral experiments have shown that the avian compass is sensitive to the inclination of the magnetic field rather than the polarity, meaning that the radical pair (RP) singlet/triplet fractional yield is expected to be invariant to field inversion. This symmetry has been seen by modeling the RP system with a time-independent spin Hamiltonian, but upon adding time-dependence to the spin Hamiltonian to model RP dynamics, the symmetry often breaks, resulting in a field inversion asymmetry (FIA). The FIA asymmetry effect can be explained by the breaking of  time-reversal symmetry due to the noncommutativity between the spin Hamiltonian at different times. However, this explanation is not informative of the relevance of the effect in biological systems. This project aims to further explore and characterize FIA in more realistic and dynamic RP systems through model system simulations. 

Samuel Dyer with Prof. Michael Martens 
 
Applications of KAN Neural Networks in Physical Problems
 
Kolmogorov-Arnold Networks (KANs) are an alternative to Multi-level perceptrons (MLPs) in applications of machine learning networks. KANs are based on the Kolmogorov-Arnold Representation Theorem, and use learnable activation functions which are modeled as splines. These learnable activation splines replace the weights of MLPs, and the nodes of the KANs sum the incoming signals. The unique quality of KANs in physical applications is their interpretability. Due to the Kolmogorov-Arnold Representation Theorem, the inner workings of the KANs are much more interpretable than MLPs, and as such, can be better pruned and optimized to address specific problems. This has been tested on Knot theory using supervised learning, which allowed the KAN to develop a new theory in regard to algebraic and geometric knot invariants. Utilizing Kolmogorov-Arnold Networks, further applications of machine learning will be assessed in various physical systems, and analyzed for efficacy. 
Nathan Fuss with Prof. Cyrus Taylor
 
Stochastic Differential Equations in Climate Change
 
It is well understood that the systems governing weather are inherently chaotic. This creates difficulties when studying the fundamental dynamics of climate change. In practice variables are typically split into two terms, a dynamic term and a “weather” term which is considered to be a stochastic variable. There are concerns that the climatic system may include tipping points – understood here to be the system containing hysteresis. These have been studied in simple models that neglect the stochastic “noise” from weather, and in large simulations which presumably take this into account but can be difficult to understand. My project will be to analyze one or more climate models for tipping points, extending the simple models to account for random noise as a basic model for the chaotic dynamics of weather. The goal will be to understand how this may change the detailed dynamics as the system approaches tipping points. I will begin with a simple model for the ice cover, and then hopefully extend this to the question of instabilities of the Atlantic Meridional Ocean Current (AMOC).

Margaret Goldstein with Prof. Johanna Nagy

Beam Mapper for CMB-S4

CMB-S4 is a new ground-based cosmic microwave background experiment that will use more than 500,000 cryogenically cooled superconducting detectors. These detectors are fabricated in arrays on silicon wafers and integrated with feedhorns and readout electronics to form a module. Before these detector modules can be deployed, they need to be tested to ensure they meet the operating requirements of CMB-S4. One important test for these detector modules is mapping their beams over the range of frequencies of their intended use. A beam mapper consisting of a thermal source mounted on multi-axis translation stages can be used to test these detector modules. The purpose of this project is to design, develop, and test a beam mapper to create an efficient method for testing CMB-S4’s detector modules.

Julia Gumina with Prof. Michael Hinczewski

Ecological Interactions of Cancer Cells in Tumors

Recently published work from the Hinczewski and Scott labs (PRX Life 2, 023010 (2024)) has found that tumors act as small ecosystems in which cancer cells interact and compete for limited resources. This project aims to expand on this discovery and create a more realistic mathematical model that takes into account the spatial distribution of cells within a tumor. The goal is to predict how the spatial distribution affects cell interactions in an evolving tumor, and ultimately how this impacts the evolution of drug resistance.

David Kuhtenia with Prof. John Ruhl and Prof. Johanna Nagy

Analyzing Metal Mesh Filters with Transmission Line Models

Millimeter-wave detectors used to measure the CMB require cryogenics to work well. In addition, other frequencies of radiation that enter the system (such as blackbody radiation from warmer parts of the system) need to be filtered out. One common type of filter is a filter where metal is deposited on a substrate dielectric, known as metal mesh filters. It is possible to use a transmission line model to analyze the different types of filters. To parametrize the metal mesh filters Ansys HFSS can be used to simulate the flow of EM waves through the filters and find the equivalent transmission line parameters. This project will be focused on modeling loss in the filters and finding values for many different filter shapes.

Austin Kuntz with Prof. Shulei Zhang

Magnetoplasmons in Altermagnets

Altermagnets are an emerging class of magnetic materials that exhibit characteristics of both ferromagnets and antiferromagnets. They feature alternating spin configurations that result in zero net magnetization, along with crystal symmetries that lead to momentum-dependent spin splitting. This project aims to understand the behavior of magnetoplasmons—collective oscillations of electron density in altermagnets influenced by their unique spin polarized electronic bands. The study will focus on the dispersion relations of both surface and bulk magnetoplasmons and their controllability. Investigating these relations is not only of fundamental interest but also has potential implications for the development of future spintronic and plasmonic devices, where controlling plasmon modes with spin current could enable new functionalities.

Caidan Moore with Dr. Todd Monson (Sandia National Laboratories)

Estimating The Dzyaloshinskii-Moriya Interaction Using Machine Learning

This century, hysteresis measurements of magnetic materials have been revolutionized by the advent of automated first order reversal curve (FORC) measurements. Bolstered by this development, significant research has been conducted on various magnetic materials. By performing a series of hysteresis measurements using FORC, the quantum interactions in a thin-film lattice can now be determined efficiently and effectively. However, decoupling quantum effects in thin film magnetic hysteresis experiments remains one of the most significant challenges in the field of FORC analysis. This project seeks to extract the antisymmetric Dzyaloshinskii-Moriya Interaction (DMI) through the use of computer modeling and machine learning (ML). The work will be supported by the RadEdge program at Sandia National Laboratories under a Department of Energy grant for spintronics research. Recently, Fugetta et al. (2023) successfully obtained the DMI of experimental films by simulating a square spin lattice undergoing a FORC measurement, then training a ML model on the constructed data. This project will be based on their model. First, a Python program will be constructed that determines the stable configuration of a spin lattice. Then, a ML model will be trained on the computational model. Finally, this measurement will be compared to experimental results to determine the accuracy of the model. If successful, additional features will be built, including the simulation of non-square lattices.

Joey Rodriguez with Prof. Stacy McGaugh

Testing a New Method for Galaxy Cluster Mass using Weak Lensing

I test a new method of inferring mass profiles of galaxy clusters from weak gravitational lensing observations devised by Tobias Mistele and Amel Durakovic. The method assumes spherical symmetry and a fairly small convergence. To test the method, I use artificially created data to apply to the method to see if I get results I would expect. I also use real data and compare inferred mass profiles of galaxy clusters to existing literature using other methods.

Edward Rowley with Prof. Benjamin Monreal

Construction and Testing of a Skipper CCD Test Stand

One of the prevailing theories seeking to explain the phenomenon of dark matter is that of weakly interacting massive particles (WIMPs), which despite being theorized to comprise 85% of matter in the universe, would interact both infrequently and with a low enough intensity regular matter so as to make detection extremely difficult. Detecting these particles is therefore a matter of building a detector apparatus which is not only large enough to observe such interaction events, but sensitive enough to distinguish an interaction event from background noise. While Germanium-based detectors have become state of the art for such experiments, a more cost-effective alternative is that of Skipper CCDs. Similar to the devices found in some digital cameras, they consist of an array of silicon capacitors, each capable of registering individual photon events in the form of an electric charge. This charge can then be transferred to a readout unit via a series of transistor gates but importantly, the charge is not destroyed upon readout, allowing for the charge from a single capacitor to be measured multiple times, thus minimizing background noise by taking the average of many readouts. Fermilab has begun to explore the potential of this approach in the Oscura project, for which they have developed custom Skipper CCDs, low threshold acquisition (LTA) readout boards, as well as proprietary software for operation of the LTA boards. The aim of this senior project is to acquire a Skipper CCD and LTA board from Fermilab, design a cryocooled test-stand apparatus for the CCD, and to setup a computer with the necessary software in order to create a new experimental apparatus for the undergraduate advanced instrumentation laboratory.

Oliver Schatzle with Prof. Shulei Zhang

Reversing of the Polarization of a Magnon Using Fixed Geometry Nano-Aperture

It is possible to excite a Magnon through a polarized current. This polarized current induces two types of Spin transfer torque, the first destabilizes the magnon state and then the second reverses it’s polarity. This ability to apply current to a magnon can be achieved by passing current through a nano-aperture, and through specific geometry choices of  the nano-aperture, the amount of current required to reach a critical current for reversing polarity can be greatly reduced. The ability to control polarity change through current as well as reducing the required critical current have many applications in data storage and quantum computing, among others.

Simon Silverstein with Prof. Johanna Nagy

Systematic Forecasting for the CMB Taurus Experiment

Ananda Smith with Prof. Glenn Starkman and Dr. Craig Copi

Gravitational waves in non-orientable universes

Modern cosmology uses general relativity to describe our Universe as a manifold with a local geometry known at any point in spacetime. However, general relativity says nothing about the global structure, or topology, of a spacetime manifold; several topologically distinct manifolds can admit the same local geometry. It is therefore worthwhile to investigate observable consequences of topologically nontrivial universes that admit one of the three local spatial geometries (Euclidean, spherical, and hyperbolic) permitted in the current standard cosmological paradigm. In this project, we aim to solve the tensor eigenmodes of the Laplacian, which describe gravitational waves, in spacetime manifolds that admit Euclidean spatial geometry and are non-orientable, meaning there exist curves in the space that transport right-handed tetrads into left-handed tetrads. Once these eigenmodes are known, we will predict the statistical properties of fluctuations in the cosmic microwave background (CMB) temperature and polarization (E and B modes) within these non-orientable spacetimes.

Anna Stacy with Dr. Jacob Scott (Cleveland Clinic)

Validating computational predictions for the evolution of drug resistance in E. coli using an automated experimental culture system

Pharmacokinetic–pharmacodynamic (PKPD) models, which describe how drug concentrations change in a patient and the drug’s effect on pathogen growth, can be employed to better understand and predict the evolution of drug resistance. In this project, we seek to design and verify optimal dosing regimens that can be employed in clinical settings to potentially minimize resistant E. coli infections. We will continue to develop a computational model that generates evolutionary predictions which can be validated with our low-cost automated experimental culture system, the EVolutionary biorEactor (EVE). Specifically, we adapt FEArS (Fast Evolution on Arbitrary Seascapes), a software package for modeling biological evolution, to predict the evolution of drug resistance in E. coli for different dosing regimens and pharmacokinetic profiles. We will also examine various levels of treatment non-adherence to optimal dosing regimens through computational and experimental methods to understand the effect of non-adherence on mutant persistence. Upon experimental validation, this probabilistic model and EVE integration can be extended to a range of pharmacokinetic profiles, dosing intervals, and treatment adherences to design optimal dosing regimens and evaluate the risks of antibiotic non-adherence. 

Evan Steirman with Prof. Stacy McGaugh

Quantifying the Baryon Deficit in Galaxies and Clusters

The cosmic baryon fraction f_b, the ratio of baryonic matter to total matter in the universe, is constrained by observations of the cosmic microwave background to be f_b = 0.16 +/- 0.01. Observations of galaxies and galaxy clusters at all scales yield two surprising and problematic results: first, the baryon fraction is lower than 0.16 everywhere, and second, the baryon fraction in an object follows a monotonic low-scatter relation with its virial mass. In this project, I will examine the upper limit of the observed baryon fraction at large scales, further quantify the baryon deficit as a function of mass, test abundance matching relations for consistency with the Schechter function, and compare the predictive capabilities of stellar and baryonic mass paradigms. Clarification of these topics will allow for a more precise understanding of the missing baryon problem and may provide insight into key points of advancement toward possible solutions.

Kai Yamagami with Prof. Svetlana Morozova 

Phase Behavior of Polyacrylamide Hydrogels in Complex Coacervate Droplets

Phase separation is a fundamental and ubiquitous organizing principle in cells and performs critical functions, including biomacromolecule sequestration and signal pathway activation. Membrane-less organelles like stress granules and nucleolus are well-studied phase-separated liquid droplets that can assemble and disassemble in response to environmental cues. Molecular crowding in cells is known to facilitate phase separation, but little is known about how elasticity, another ubiquitous character of cytosols and the extracellular matrix, affects phase separation. Here, we will use complex coacervates as a model of phase-separated droplets and encapsulate them in hydrogels of tunable modulus to study the elastic effects on phase separation. Elastic effects on phase separation are experimentally observed and theoretically validated in a high-surface tension system where oil droplets are encapsulated in silicon gels. In this case, oil droplets deform the gel beyond the mesh and grow until the equilibrium size scale. In cells, however, there is little surface tension between cytosols and membrane-less organelles. Hydrogels and complex coacervates also have low surface tension. Because the gel mesh size is in nanometer order while the droplet is in the micrometer order, surface-tension-supported deformation beyond the gel mesh size is unlikely. I propose an alternative hypothesis that gels swell or de-swell locally around the complex coacervate droplets by the balance of thermodynamics of the constituents: gels, polymers, salt, and water. The hypothesis will be tested by simultaneously imaging gels and complex coacervate droplets using confocal microscopy. 

Elizabeth Zhou with Prof. Giuseppe Strangi

Optimization of Tunability in Metalenses using Liquid Crystals

A metalens is a flat, ultrathin lens that uses nanostructures to focus light, offering a compact alternative to conventional optics. However, the limitations of modifications in metalens post fabrication are a well-known stumbling block in expanding its use in real-world applications. To overcome this limitation, the infiltration of the lens with a nematic liquid crystal has been proposed. By changing between the nematic and isotropic phases of the liquid crystal, a focal distance shift of 80μm is observed. Control over the orientation of the liquid crystal (planar/hometropic) in relation to the metalens structure will allow for greater tunability. In order to further optimize this tunability as well as the focal capabilities of the lens itself, this project will design and fabricate metasurfaces using the Two-Photon-Polymerization (TPP) technique.
Theodore Zhuge with Prof. Mike Hinczewski
 
Enhancing AI Painting Attribution:  Separating Paint from Canvas in Oil Paint Artwork
 
This project integrates artificial intelligence with art analysis to identify which parts of a painting were created by different individuals. While current models, using optical scans, can distinguish the contributions of various artists within a single painting, the AI encounters difficulties when the canvas conditions vary from those it was trained on. The way brushstrokes interact with different canvas textures affects the AI’s performance. To address this issue, the research focuses on generating digital simulations of oil paint, enabling the AI to be trained to separate the effects of canvas texture from the painted brushstrokes. The goal is to train the AI to consistently and accurately attribute sections of a painting, regardless of the canvas used. This digital approach offers faster and more extensive data resources, promising to improve the AI’s analytical capabilities.