High Entropy Alloys (HEA) are alloys that contain five or more elements in equal or near equal atomic percent. This compositional complexity produces relatively simple solid-solution structures. Specifically, a Eutectic High Entropy Alloy (EHEA) is an HEA where BCC and FCC phases exist in an eutectic reaction. This dual phase FCC/BCC microstructure enables favorable strength and ductility, a combination that cannot be found in traditional alloys. AlCoCrFeNi2.1 is an EHEA that has shown promising results for microstructural evolution through strain partitioning and twin-assisted plasticity paired with recrystallization annealing. Due to the recency of EHEA, a practical gap remains in connecting manufacturing-style processing to phase evolution, specifically identifying how strain hardening and recrystallization annealing induces new intermetallic phases in AlCoCrFeNi2.1 EHEA and how the microstructural changes impact key properties. The objective of this research was to observe and quantify the microstructural changes produced in AlCoCrFeNi2.1 Eutectic High Entropy Alloy after cold rolling and annealing. This was done by cold rolling and annealing AlCoCrFeNi2.1 under different cold rolling thickness reductions and annealing temperatures and then analyzing the microstructural evolution using  X-Ray Diffraction and Electron Backscatter Diffraction (EBSD).

This research project examines how aging classroom infrastructure and poor environmental design can negatively affect student learning and mental well-being at New Mexico Tech. Many academic buildings contain classrooms with limited natural light, minimal aesthetic design, and environmental conditions that may not support optimal learning. Research in environmental psychology and educational design shows that factors such as lighting, air quality, acoustics, and access to natural elements can influence cognitive performance, concentration, and emotional responses in learning environments. Studies on classroom design and built environments indicate that uncomfortable or poorly designed spaces can reduce comprehension, increase distraction, and negatively affect students’ perceptions of their learning experience. This project proposes investigating improvements to classroom environments through design strategies that support comfort and psychological well-being. Potential approaches include increasing access to natural light, incorporating biophilic design elements such as plants, and improving the overall sensory quality of classroom spaces. These strategies have been shown in prior research to improve perceived environmental quality, productivity, and student engagement. Addressing the physical learning environment is significant because students spend substantial time in these spaces, and improving classroom conditions may enhance both academic performance and overall student well-being. 

At New Mexico Tech, the lack of healthy alternatives at Chartwells Dining Hall presents a significant barrier to student wellness. Utilizing a secondary analysis of nutritional literature, this research identifies a correlation between consumption of unhealthy foods and adverse outcomes, including increased mental distress and lower academic achievement. Findings suggest this environment places students at an academic disadvantage. To address this, we are advocating for improving the nutritional quality of Chartwells' offerings while increasing student awareness and use of food assistance programs such as S.N.A.P. Enhancing these standards and support systems is essential to fulfill the university's ethical responsibility and ensure that campus resources effectively support student health and academic success. 

The developed ground-penetrating radar (GPR) system is designed to be a compact impulse radar transmission device that performs analysis on subsurface reflections. The system architecture centers on the generation of a 400 MHz impulse waveform transmitted through orthogonally polarized bowtie antennas. The reflected signals are then received and digitized for analysis. Power distribution is managed through regulated 12 V and 5 V rails with subsystem protection and removable battery operation to support portability. An FPGA performs high-speed signal generation and real-time preprocessing of received data, while a single-board computer provides system-level control, configuration, and user interfacing. 

Bastnasite (REECO3F) is a major ore mineral for light rare earth elements (REE) such as Cerium (Ce), Lanthanum (La), Neodymium (Nd), and Yttrium (Y). These elements play a crucial part in high-tech industries such as renewable energy, permanent magnets, and catalytic converters. Bastnasite forms a solid solution with the mineral hydroxylbastnasite (REECO3OH) in which a hydroxyl group (OH- ) replaces the F. In this study, we characterized synthetic Ce- and La- bearing bastnasite using Raman Spectroscopy, Scanning Electron Microscopy-Energy Dispersive Spectroscopy (SEM-EDS), and X-Ray Diffraction Spectroscopy. The synthetic bastnasites characterized include pure bastnastine-Ce and bastnastie-La end members as well as mixed REE bastnasites ranging from 90% Ce and 10% La to 10% Ce and 90% La. The atomic proportions of Ce, La, and F in the bastnasites were measured using SEM-EDS and used to determine REE/F and Ce/La ratios. Raman spectroscopic analyses indicate partial replacement of the F with OH- groups in the bastnasites. Characterization of the synthetic bastnasite is crucial for using it in further research to determine the thermodynamic properties of bastnasite. 

Biomimicry involves taking inspiration from nature to help improve human-made mechanical design. By looking to the millions of years of evolutionary trial and error, engineers are able to copy highly efficient solutions to natural problems. Robots often struggle to handle complex locomotion so we can turn to frogs, who specialize in movement. Frog legs act as springs able to store elastic energy within their tendons to quickly propel them many times their body length. Taking elements from these creatures is crucial in creating mechanical design capable of multi-terrain movement, whether it be water or mud. Scissor multi-pivot linkage systems proved extraordinarily effective in creating convincing locomotion of frog hind legs. This system offered rapid extension and complete contraction, crucial to mimicking a frog’s small profile. The scissor linkages excelled in creating movement with the extreme torque limitations due to the inherent necessity of small motors. An RC design was utilized with a controller and receiver to pilot the robot from range. To encapsulate the entirety, a small carriage was created to hold the entirety of the robot that anchored the legs without limiting their extension. Extreme importance was placed upon reducing size, weight, and part count while maintaining the visual silhouette of a frog. While we could successfully mimic visual and mechanical aspects of frog movement it is impossible to capture the true efficiency of million year old biological systems. The results of this research demonstrate the usefulness of biomimicry to enhance future biomimetic systems. 

Lava caves form when surface lava cools and solidifies while molten lava beneath drains away, leaving hollow tunnels of hardened basalt. These subterranean ecosystems are characterized by stable microclimates, low light levels, and low nutrient availability limiting the number of organisms that can survive. Cave fungi are likely heavily influenced by particulate organic matter (e.g., leaf litter) or animal activity (e.g., bat guano), but it has also been suggested that fungi can form microbial associations with bacteria to enhance nutrient extraction from cave substrate. This project aims to characterize the bio- and chemodiversity of a previously overlooked lava tube cave near Grants, New Mexico. We will sample several isolated regions within the lava cave both by swabbing and taking small rock chips. Targeted sampling will include soda straw and lava stalactite features to assess the bio-chemodiversity associated with secondary mineral formations. Fungi will be cultured on low nutrient media and subsequently isolated in pure culture. We will identify representatives of unique morphotypes through DNA extraction, polymerase chain reaction (PCR), and Sanger sequencing. Crude extracts containing fungal secondary metabolites will be tested against bacterial and fungal pathogens to screen for antimicrobial activity. Promising isolates will undergo fractionation of chemical components for further characterization. This work aims to establish a foundational understanding of previously undocumented geomycological communities and broaden the understanding of biological and chemical diversity within volcanic subterranean ecosystems. 

Plasma is created by subjecting a neutral gas to a large enough applied electric field, after which the gas consists of free electrons and ions. Inherent to plasmas, is the emission of photons ranging from the extreme ultraviolet (EUV) to the far-infrared (IR). Plasma photons interact differently with their surroundings based on the pressure of the environment as well as the plasma composition. Only photons in the EUV regime are energetic enough to cause direct ionization of many industry relevant gases but are easily absorbed at higher pressures. To study this, an experiment is being designed to enable observations of photons with wavelengths shorter than 100 nm while still allowing observation up to 800 nm. The experimental set up consists of plasma generated in a high pressure chamber (10’s of mTorr) and a spectrometer kept at high vacuum (< 10-6 Torr) separated by a series of pin holes. To effectively measure light emission, a differential pressure must be established between the two systems where photons that are generated via the plasma at higher pressure are quickly transported to a lower pressure region where they are not absorbed. To model the pressure differential, a Stochastic PArallel Rarefied-gas Time-accurate Analyzer (SPARTA) simulation based upon the Direct Simulation Monte Carlo (DSMC) method was used to simulate how N2 and O2 particles interact with each other as they move from different pressures in a closed chamber. 

Radio frequency plasma discharges are an important tool in many industrial processes such as the fabrication of microelectronics. Thus, detailed characterization of their behavior is key to improving the capability of modern manufacturing techniques. Light emission, which is ubiquitous with any plasma discharge, may influence plasma properties via interaction between plasma species and photons, or through light-material interactions. Here, a capacitively coupled plasma (CPP) is generated between 2 cylindrical electrodes with a gap spacing of 3 cm in Argon to study spatially resolved light emission. An RF generator providing a 20 W 13.56 MHz signal with an average reflectant of 2 ± 1 W ignites and maintains the plasma. Visually, it is observed that there are two intensity maxima peaks near the electrodes at 176mTorr. Conversely, only a single intensity peak is observed in the middle of the gap when the pressure is decreased to ~76mTorr. This light emission is quantified using spatially-resolved optical emission spectroscopy (OES). Light emission from the plasma is directed into an Andor Kymera 328i spectrometer using a pair of plano-convex lenses and detected on a Hamamatsu photomultiplier tube. Both total intensity and individual line intensities are examined as a function of distance away from the cathode. The speculation is that there is a transition from γ-mode (observed double peak intensity) to α-mode (single peak intensity) within the gap based on the dependence of the pressure. 

The research focused on investigating the reason behind student cheating via large language models (e.g., ChatGPT) and ways of reducing cheating in this way. Students have been turning to language models more due to the models ease of use and availability. Students who use these models for correcting gaps in knowledge not only may result in students absorbing incorrect information, but doing so also robs them of the chance to perform additional research outside of a classroom setting, a critical workplace skill. The main goal of the research was on compiling other studies to identify some of the causes of student cheating, and use them to find feasible solutions. The solution that was most supported and feasible would be a combination of making resources for higher level classes more accessible, and focusing on real world applications of the material where possible. 

Continuing current (CC) in lightning occurs when the current in the lightning channel flows continuously for longer than in a typical lightning discharge. Lightning with CC neutralizes more charge and generates more heat than lightning without CC, making it more likely to cause fires and other damage. As a result, understanding CC is an important area of lightning research. While negative cloud-to-ground lightning, originating in the negatively charged region of the cloud, is more common, positive cloud-to-ground lightning, originating in the positively charged region of the cloud, more often has CC. In this project, lightning data were collected using fast and slow electric-field antennas and Lightning Imaging for Ground-truth and Electric-field Recorder (LIGER) cameras, both located at Workman Center. There are two LIGER stations, one recording at 120 fps and one at 240 fps, each with two cameras, providing a wider field of view. Data from these instruments were merged with data from the Langmuir Lightning Mapping Array. They were then analyzed alongside Earth Networks data to assess correlations between CC duration and other factors, including peak current, charge layer altitude, and the area covered by lightning during the CC. Preliminary results indicate a positive correlation between CC duration and lightning area extension during that time. Positive CC tends to have a higher peak current compared to negative CC. This work could be expanded by building a more statistically robust dataset, enabling a deeper understanding of how CC in lightning correlates with other factors. 

Pharmaceutical Medicine comes with different sizes, packaging format and weight. Handling, sorting and storage of pharmaceutical medicine is quite challenging because it requires careful orientation control, precise positioning and contamination free handling. Although current practices follow strict precautionary guidelines, Minor misalignment or handling error or medicine can cause damage, labeling error and cross-contamination.

The current solution falls into mainly two categories- Manual Handling and Fully automated fixed systems. The manual handling process is flexible but it is more error-prone and depends on human fatigue and cognitive load. Also it is not scalable as human physical works capabilities are limited. Similarly, fully autonomous handling systems are very efficient, but it requires a fixed environment, relies on a pre-defined path and needs fixed rack geometry. The fully autonomous robotic handling systems become inefficient and challenging when racks are in cluttered condition, require handling of different size packages and when the racks layout changes very frequently. It was highlighted in the collaborative robotic research that unstructured environments reduce the efficiency and reliability of fully automated pre-programmed robots. Also it was mentioned in medical oriented research that robots enhance human capability when working together, which is known as Human Robot Collaboration (HRC). Therefore, a semi-assisted, human-in-loop framework is required to eliminate the existing limitations in pharmaceutical medicine handling. The framework combines depth-based perception, structured motion planning, real-time teleoperation. Overall, the research proposes a ROS2 based vision guided teleoperation framework for pharmaceutical medicine shorting and handling using a 6DOF robotic manipulator. 

Digital twins (DTs) are increasingly utilized in healthcare to provide virtual representations of the human body. However, many models lack a focus on physical phenomena, such as motion and structural strain. The Physical Health Digital Twin (PHDT) addresses this gap by utilizing wearable, self-powered mechano-luminescent-optoelectronic (MLO) sensors to continuously map strain during human movement. To effectively bridge the gap between this physical sensing hardware and its digital counterpart, a comprehensive data processing and visualization pipeline was developed. The experimental plan involved creating custom Python-based software to synchronize and cross-correlate direct current (DC) signals from the MLO sensors with reference to joint kinematics extracted from video recordings. To interpret this data, an interactive graphical user interface (GUI) was engineered using Tkinter and OpenCV. This tool dynamically reconstructs a user's joint kinematics, in this case knee angles, in a virtual environment. The principal finding of this development is a visualizer that seamlessly synchronizes the digital skeletal reconstruction alongside real-world reference video playback and live-plotted strain data. This allows researchers to visually fine-tune time alignments and export composite analytical videos. Ultimately, these software tools enable the qualitative and quantitative validation of the MLO sensor. By creating a reliable interface between physical sensors and digital models, this framework establishes a robust computational foundation for next-generation personalized healthcare monitoring and physical rehabilitation technologies. 

Traumatic brain injuries (TBIs) affect thousands of people annually in military, sports, and civilian populations. However, TBIs are under-diagnosed and under-investigated due to challenges in monitoring in vivo brain tissue mechanics during head insults. A critical damage location is the gray-white matter interface due to the mechanical stiffness difference between tissues. Utilizing a custom shear device, we deformed ex vivo bovine tissue to identify strain concentrations at the gray-white matter interface and reproduce it with multi-layer hydrogels sharing a cross-linked interface. Strain and strain rate measured with digital image correlation at a frame rate of 10,000 fps demonstrated replication of known damage-inducing parameters. To quantify cellular damage at these interfaces, a cerebral organoid-on-a-chip (COCh) was fabricated. These models consist of a polyacrylamide hydrogel, which mechanically mimics brain tissue, with embedded cerebral organoids. A multi-layer gray-white matter COCh was fabricated to place a cerebral organoid under deformation at interfaces and to quantify damage for comparison to a single-layer COCh model. Under high loading conditions, single-layer COCh models show that embedded organoids experience strains and strain rates capable of inducing cellular damage, whereas the multi-layer COCh did not due to limited interactions at the organoid-hydrogel interface. Ultimately, development and characterization of the COCh under TBI-like loading conditions may allow injury threshold identification by correlating strains and strain rates in organoids with biological response. Finally, these chips can be placed into full-scale brain phantoms and exposed to realistic TBI scenarios informing equipment design and preventative training policies in military and sports applications. 

What if an aircraft wing could be designed not through conventional iterative design processes, but by intelligently learning what works best, much like nature already has? Inspired by the Arctic tern, one of the most efficient long-distance flyers, this study presents a machine learning-driven approach to designing a high-efficiency unmanned aerial vehicle (UAV) wing with the goal of increasing flight time. A diverse set of wing geometries was generated and evaluated using a fast aerodynamic tool (Flow5), focusing on lift-to-drag performance across realistic flight conditions. The resulting dataset was used to train a surrogate model capable of predicting aerodynamic efficiency without repeatedly performing simulations. An evolutionary algorithm, inspired by biological evolution through selection and mutation of candidate (“species,”) was then employed to explore thousands of design variations and identify an optimal configuration that yield the maximum aerodynamic efficiency. To improve predictive accuracy of the surrogate model, additional samples were introduced near the optimum, reducing model error from 35% to approximately 2%. The final design was validated using higher-fidelity simulations, confirming improved aerodynamic performance. This work demonstrates how bio-inspiration, machine learning, and evolutionary optimization can accelerate UAV design while achieving superior efficiency and extended flight time. 

This project presents the design and simulation of a vision-guided robotic arm system for automated crop harvesting in greenhouse environments. While greenhouses improve agricultural productivity through controlled conditions, many key tasks such as harvesting and sorting still rely heavily on manual labor, leading to inefficiencies, inconsistencies and increased operational costs. This work addresses these challenges by proposing an integrated robotic solution that combines perception, motion planning and adaptive manipulation. The proposed system utilizes an RGB-D camera and image-based detection techniques to identify and localize ripe fruits in three-dimensional space. A robotic manipulator modeled with standard kinematics executes smooth and collision-free trajectories using optimized planning techniques. To ensure safe interaction with delicate crops, the system incorporates controlled grasping and cutting mechanisms, supported by force feedback and continuous visual feedback during operation. The entire architecture is implemented within a ROS2 framework and validated in a Gazebo simulation environment allowing for realistic testing and iterative improvement. Performance is evaluated based on detection accuracy, harvesting success rate, motion smoothness and operational efficiency. The system is expected to achieve reliable fruit detection, stable manipulation and a harvesting success rate exceeding 80 percent while maintaining cycle times suitable for practical deployment. By integrating perception, planning and control into a unified framework, this work aims to reduce dependence on manual labor, improve harvesting consistency and contribute to the advancement of scalable and sustainable greenhouse automation systems. 

Several radioisotopes of bromine are used in nuclear medicine to image and treat various diseases. This includes bromine-77 (77Br), an Auger-electron emitter used for targeted radionuclide therapy against several diseases, such as breast cancer. The traditional method of production uses an incident proton beam from a cyclotron onto a target of selenium to induce a nuclear reaction that produces these isotopes. However, cyclotrons can often be expensive and their yields are often limited by practical heat considerations, leading to the exploration of alternative pathways for the production of many radioisotopes. One alternative pathway uses bremsstrahlung radiation, where gamma rays are created by the “braking” of electrons colliding with a nucleus when traveling through a material. This research project focuses on bremsstrahlung-induced photonuclear reactions on natural bromine targets to produce bromine-77. This production route utilizes the 79Br(𝛾, 2n)77Br production route, which has little to no data previously published. Fluktuierende Kaskade (FLUKA), a Monte Carlo simulation code, and Microsoft Excel were used to model the pathway as a preliminary result to assess expectations. Following this, potassium bromide targets were prepared and brought to the Idaho Accelerator Center where a bremsstrahlung beam was used to irradiate the targets. Gamma spectroscopy with high-purity germanium detectors were used to measure the yields shortly after irradiation, and further work will be done to determine the yields and cross sections. This work is supported by the U.S. Department of Energy under grant number DE-SC0023665. 

This project investigates the feasibility of a fully autonomous robotic system for order fulfillment in low-light warehouse environments. The central hypothesis is that a LiDAR-based robotic platform can reliably perform end-to-end warehouse tasks including navigation, product retrieval, packaging, and delivery without human intervention, while maintaining efficiency comparable to traditional systems. The system is designed as a modular architecture using ROS nodes to ensure scalability and maintainability, and all development and testing are conducted initially in Gazebo simulation prior to potential hardware deployment. Based on current design and simulation parameters, the system is expected to successfully demonstrate autonomous task completion under low-light conditions. This research contributes to the fields of robotics and industrial automation by exploring a scalable framework for autonomous warehouse operations, providing a foundation for future experimental validation and hardware implementation. 

In this study, we propose a stretchable and self-powered strain-sensing strip with multi-sensor nodes to form a sensor network in clothes-type health-monitoring wearables. The sensing strip is designed using the multifunctional mechano-luminescence-optoelectronic (MLO) composites design platform, where two functional constituents are assembled to generate strain-varying direct current (DC) through mechanical-radiant-electrical (MRE) energy conversion. First, mechano-luminescent (ML) micro-composites are designed using the strain-amplifying mechanical metamaterial (SAMM) framework and fabricated by mold-casting a zinc sulfide (ZnS)-based elastomer into an ML strip. Second, mechano-optoelectronic (MO) thin films are deposited by air-brushing poly(3-hexylthiophene) (P3HT)-based thin films on the ML strip, which is followed by wiring for DC measurement and encasing with silicone-based elastomer. Last, the fabricated MLO sensing strip is validated by measuring the DC output from each sensor node when subjected to various loading cases. 

Dual jet heat transfer is characterized by a Nusselt distribution different from single jet flow. In particular, the downstream distance of the local Nusselt minimum is affected by varying the total mass flow rate and velocity ratio. Paula J. Murphy’s 2D dual jet test setup was simulated in FDS to validate the experimental flow’s Nusselt behaviors when changing both parameters. The Nusselt distribution across the flat plate with constant heat flux was calculated for nine studies after developing a structured cell mesh, setting the correct boundary conditions, and measuring the surface temperature across the flat plate. The simulations exhibited all three behaviors listed in Murphy’s paper. Likewise, the percentage errors calculated were within 15% of the experimental values and were therefore acceptable. 

Mechano-luminescent–optoelectronic (MLO) systems are designed as self-powered devices that convert strain into electrical energy. Within these systems, mechano-optoelectronic (MO) components based on poly(3-hexylthiophene) (P3HT) thin films enable the conversion of light into direct current (DC). Despite this capability, the long-term performance of these devices is limited by degradation, leading to reduced electrical output and decreased reliability over time. This study focuses specifically on the MO component to better understand the mechanisms responsible for degradation in MLO DC energy conversion.

It is hypothesized that prolonged exposure to light alters the active material and its interfaces, reducing charge generation and transport efficiency. To examine this, MO devices are fabricated by spin-coating P3HT thin films onto various substrates and exposed to controlled illumination conditions for extended durations. The resulting DC current (DCI) and voltage (DCV) are monitored over time to evaluate performance loss and stability. In addition, device multiplexing is investigated as a method for increasing and tuning electrical output, while the influence of substrate properties on thin-film quality is also considered.

Results show that electrical output decreases with increased light exposure, and further degradation occurs over time even without illumination. Multiplexing leads to increases in both DCI and DCV, demonstrating scalability. These findings provide insight into stability limitations and support the development of more durable self-powered MLO systems. 

Cosmic rays from space constantly interact with Earth’s atmosphere and create showers of secondary particles that contribute to background radiation. As a result, the particle flux increases as you go higher in the atmosphere and reaches a peak called the Regener-Pfotzer maximum before decreasing again. This study investigates how radiation counts change with altitude and whether this peak can be observed during a high-altitude balloon flight. To study this, we launched a weather balloon carrying a Radiacode 103 radiation detector. The detector recorded radiation counts throughout the flight, while the balloon rose through the atmosphere. By combining these measurements with altitude data from the flight profile, we construct a graph of radiation count versus altitude curve. We expect to observe relatively high background radiation counts closer to the surface due to radioactive mineral deposits present in New Mexico’s geology. As the balloon ascends and surface radiation becomes negligible, we expect to see a transition to a trend of increasing radiation counts with altitude, since incoming cosmic rays have to travel through less atmospheric shielding. This project demonstrates how a student-built balloon platform and Radiacode instrumentation can be used to investigate cosmic radiation in Earth’s atmosphere. The results will help illustrate how cosmic rays interact with the atmosphere and provide a hands-on example of particle astrophysics using student-built instrumentation. 

Earthquakes often appear sudden, yet they reflect the culmination of a complex buildup of stress, frictional resistance, and microscopic damage. This study asks whether artificial intelligence can reveal the physical laws governing fault slip. We use data from triaxial shear experiments on granite samples containing a precut fault inclined at 30 degrees to the loading direction. In these experiments, as loading progresses, the fault repeatedly fails through stick-slip events, generating laboratory earthquakes with varying stress drops. Each slip episode releases seismic energy and produces acoustic emission activity, providing insight into the evolving mechanics of frictional instability. Rather than using artificial intelligence as a predictive black box, we apply symbolic regression to identify compact mathematical relationships between stress drop magnitudes and acoustic emission signals. The aim is to move beyond pattern recognition toward interpretable physical discovery by identifying governing combinations of stress, deformation, frictional response, and acoustic activity that explain variability in slip behavior. This approach enables the identification of scaling laws and mechanistic structure that are difficult to derive from theory alone. The broader significance extends beyond laboratory systems. By extracting physically meaningful equations from controlled fault slip experiments, this work offers a pathway toward improved understanding of earthquake nucleation, seismic energy release, and frictional failure in brittle rocks. It demonstrates how data-driven approaches can uncover governing physics rather than merely describe observations. 

Underground mine pillars are subjected to sustained and cyclic stresses that can induce time-dependent deformation, progressive damage, and eventual instability. These responses are associated with creep, fatigue, and alternating creep-fatigue loading, yet their relative effects on failure time, damage accumulation at failure, and crack-evolution stages remain poorly understood. This study compares sandstone behavior under creep, fatigue, and alternating creep-fatigue loading to determine which condition causes the earliest failure, which results in the greatest damage at failure, and whether cracking stages can be identified and classified using unsupervised clustering. Sandstone specimens were tested under these loading conditions with continuous acoustic emission (AE) monitoring. Mechanical response, AE parameters, and the slope of the frequency–magnitude distribution (b-value) were analyzed. Subsequently, AE waveform spectrograms and convolutional neural network (CNN) autoencoders were used to extract latent-space features, and K-means clustering was applied to identify fracture patterns. Results show that fatigue caused the earliest failure and produced the highest AE activity, indicating the most intense rock damage. K-means clustering also revealed three distinct clusters for each loading condition, interpreted as initial compaction, subcritical crack growth, and critical crack growth, enabling objective discrimination of damage states. These findings demonstrate that the loading path strongly controls time-dependent damage evolution and microseismic response in rock, underscoring the importance of incorporating creep-fatigue interactions into long-term stability assessments of underground excavations. 

Jupiter provides a natural laboratory for studying fluid dynamics at scales unavailable on Earth, with vortices that can last from days to decades, and trace atmospheric motion and energy transport within the planet’s upper troposphere that provide information about how these vortices interact with their environment. We develop and apply an automated ellipse-fitting algorithm to detect, track, and quantify the changes in vortex shape and motion over time. This approach allows the measurement of potential oscillatory behavior of ellipse properties (e.g., position, size, and orientation) over time. We analyze global maps from the 2001–2002 Cassini flyby of Jupiter, applying contrast enhancement, thresholding, and contour-based ellipse fitting to identify vortex features. For each vortex, the algorithm records center coordinates, semi-major and semi-minor axes, and position angle frame by frame to measure drift rates and shape variability. Time series are detrended and normalized to look for periodicities using power spectra analysis, as has been done with larger spots like the Great Red Spot. Measurements from the fitted properties of four vortices in the South-South Temperate zone show that the position of the center, the orientation angle, and the semi-major and semi-minor axes all show roughly linear trends with time. Detrended residuals demonstrate whether there is potential oscillatory behavior. This method identifies and tracks vortices and provides a framework for long-duration atmospheric studies. Higher-cadence datasets can improve the detection of oscillatory periods and allow us to gain a better understanding of how vortices are interacting with their surroundings. 

Studying soil erosion in various landscapes helps us better understand how surface contaminants are transported and how long it takes them to impact waterways. Although mine tailings and deposits of historical nuclear tests pose potential danger of contaminant transport to vital areas, less is known about the sediment transport rate in the deserts of central New Mexico. Erosion is the process by which soil and rock particles are broken down and transported by varying transport agents such as wind, precipitation or human and animal activity. Our field area is located in Socorro, NM, about 4.5 miles northwest of the New Mexico Tech campus. We set up a total of 13 monitoring sites located at different slope angles on the south facing hillside. The results show that smaller tracer grains (~0.5mm - 1mm) had greater average transport distances and rates than the larger grains (2mm). We also observed that sites with greater slope angles had larger average transport distances and rates for both particle sizes compared to smaller slope angles. These results are compatible with other studies such as one done in the Sierra Nevada mountain range of California. We are also exploring how animal activity contributes to these patterns and impacts the movement of contaminants. Based on our current data, contaminants of this size take from about 100-200 years to be transported into ephemeral creeks, which can negatively impact the organisms that rely on them. These estimates help to understand the mobility of contaminants and plan possible remediation efforts.