Advisor(s): Peter N. Saeta.

Team: Eric Daniel Thompson-Martin ’22, Mary Victoria Anderson ’21, Hannah Lucia Davalos ’21, Chaitanya (Chai) Dasharathi Karamchedu ’21, and Samuel (Sam) Dana Ness ’21.

A lipid bilayer separates a cell from its surroundings and determines what substances, such as therapeutic drugs, may enter the cell. Because in situ research on cell membranes is difficult and drug development is costly,  research on the behavior of proteins embedded in lipid bilayers is often done with molecular dynamics simulations. This project has studied the output of a continuum model recently developed at LLNL and found that the lipid concentrations therein can be described by a multivariate gaussian model. This model powers an application, GRuMPy (Generative Membranes in Python), that allows researchers to construct, simulate, and analyze realistic membranes starting from incomplete specifications. As part of this process, we developed and performed two validation tests to ensure that there was appropriate probability density in the generative model over compositions that are reasonable in the continuum model. We also fit and tested an asymmetry model that allows the application to take percent compositions as input and output full membranes with appropriate proportions of lipids in the inner and outer leaflets. In addition, we provide a tool that allows researchers to train and deploy generative models based on new or updated continuum model output.

Given that we generate membranes from a model, and that GRuMPy provides the ability to simulate the compositions that it generates, we also sought to validate the generated compositions as reasonable. Validation tests done on membrane simulations of size 5 nm to 30 nm using compositions generated by the application show that the membranes generated have simulated properties identical to membranes in the continuum model. Compositions generated by the application show no significant deviation from the continuum model in area per lipid, bilayer thickness, area compressibility, or order parameter. In effect, with the assumption that the continuum model data is composed of reasonable membrane compositions, GRuMPy is able to generate physiologically realistic membranes.

Advisor(s): Peter N. Saeta.

Team: Emma Frances Cuddy ’21, Rebecca Fei Qin ’21, Rakia Segev ’21, Eric McLaughlin Weiner ’21, and Rachel Emily Cohen ’21.

Scientists at Los Alamos National Laboratory use an optical interferometric technique to observe moving surfaces with high time resolution. Light from a signal laser travels through an optical fiber before shining on the surface under test. The small portion of the beam that reflects from the surface and re-enters the fiber is mixed with a reference laser signal and the combined signal as detected by a fast optical detector is digitized at rates up to 50 GHz. The resulting data stream must then be analyzed to identify signals that encode the motion of the surface. The velocity of the surface along the direction of the laser beam produces a proportional frequency shift in the reflected light; these frequency shifts can be determined from a spectrogram of signal intensity as a function of both frequency and time calculated from the ( V(t) ) data using standard fast Fourier transform techniques. 

The reflected light is often weak and may vary in intensity over the short time of the experiment, which makes separating the signal from background noise difficult. Some experiments may cause the surface to eject particles, which can efficiently scatter light back to the fiber. In such cases many different velocities may be present in the spectrogram. Los Alamos seeks a more automated way to process photon Doppler velocimetry (PDV) files containing ( V(t) ) data to produce surface velocity and uncertainty as functions of time, where such surfaces exists, and to quantify velocity distributions for ejecta.

Advisor(s): Peter N. Saeta.

Team: Isabella (Isabel) Yibei Duan ’22, Nicholas A Koskelo ’20, and Rikki M Walters ’20.

HRL is currently developing a qubit built from three interacting quantum dots. Although tuning up qubits this way is difficult, we have begun to develop a machine-learning model to automate this process. Successfully tuning up the qubit requires the model to adjust voltages on six gates to load a single electron into each of three closely spaced quantum dots. To reduce the time to compute charge occupancy plots during training, we simplified the problem to analyze a two-dot system. Our model performs well under ideal conditions, but when factors like thermal noise are included the convolutional neural network struggles to glean useful features from the simulator’s dot occupancy plots. The model managed to correctly tune up the dots 74% of the time in the absence of noise, but was only able to tune up the dots 38% of the time and 16% of the time with cold and hot noise, respectively. The results suggest that our deep reinforcement learning model on its own will not be able to properly tune up the quantum dots, but we have outlined steps for improvement such as the addition of a classifier and an improved simulator.

Advisor(s): Peter N. Saeta.

Team: Corbin J. Bethurem ’19, Evan Joseph Hubinger ’19, John Alexander Jeang ’19, and Vivian Ngoc Thuy Vy Phun ’19.

Cellular membranes are the barrier protecting the inner cell components from the outside environment. They are composed of proteins and lipids that form lipid bilayers of varying complexity. One approach to understanding the behavior of lipid bilayers uses molecular dynamics simulations of realistic size and complexity. Realistic bilayers are asymmetric, having different concentrations of various lipids in the two leaflets. This asymmetry affects the stability, sometimes approximated as “flatness,” of the simulated bilayer. Many properties have been used to judge whether simulated bilayers are stable and realistic, but a quantitative definition incorporating multiple properties has not been proposed.

For the Lawrence Livermore National Lab Clinic project, we will create and analyze lipid membrane simulations to provide a deeper understanding of how asymmetries of varying complexity within the lipid bilayer leaflets affect membrane properties. These properties may include area per lipid, area compressibility, order parameters, and bilayer thickness, among others. From these properties, we will attempt to create a metric of stability for asymmetric bilayers. We will also develop a streamlined process for simulating and analyzing stable bilayer membranes. The final product will be used to inform and benefit future research in membrane biophysics.

Advisor(s): Peter N. Saeta.

Team: Madison Rae Blumer ’19, Sophia (Sophie) Laurice Harris ’19, Mengzhe Li ’20, Luis Angel Martinez ’19, and Michael Untereiner ’19.

The Claremont Locally Grown Power (CLGP) Clinic Team at ������Ƶ is analyzing new photovoltaic technology, invented by idealPV®, to evaluate its efficiency. This report presents a status update for three distinct aspects of the project: (1) a field study evaluating the performance of idealPV’s technology versus the industry standard; (2) a model, based on data taken in the field study, to predict watt-hour production of both types of solar arrays, with inputs of insolation, ambient temperature, system panel architecture, and sun angle; and (3) documenting the conditions and effects of reverse bias on photovoltaic cells. The team has secured a site for the field study and researched the relationship between temperature and power output of a solar cell for the model. Additionally, they have made a rudimentary model using data from PVWatts. The team completed the analysis documenting the effects of reverse bias on a solar cell and delivered their analysis to the liaisons. Sections one through four of this report detail the work that was completed by the team in the Fall of 2017. The last section of the report details project management for the upcoming Spring 2018 semester.

Advisor(s): Thomas D. Donnelly, Richard Campbell Haskell, Peter N. Saeta, and Qimin Yang.

Team: Florence Joan Walsh ’20 and Quentin Barth ’19.

This project aims to measure the size dependence of the dielectric constant of barium titanate (BTO) nanoparticles in epoxy composites. To accomplish this, we developed improvements to an existing sample fabrication process to reduce defects within the composite and improve yield, as well as improved computational models of particle size, shape, and agglomeration. In previous years, HMC Sandia Clinic teams used a ball milling procedure to reduce particle agglomeration and fabricate composites containing nanoparticles with diameters ranging from 200-nm to 500-nm. However, the dielectric constants for many of these samples did not match the predictions made by computational models that assumed 0% particle agglomeration. We examined the particles at various stages throughout the manufacturing process using DLS and SEM, and used image analysis techniques to extract information on particle size, shape, and agglomeration from microscopy images. This information was then used to inform finite-element computational models. Additionally, we employed a new, low-viscosity epoxy and a rotary evaporator to improve particle dispersion and manufacturing yield. Next semester, we will use these improvements to fabricate composites of nanoparticles with diameters ranging from 50-nm to 500-nm at both 10-vol% and 20-vol% loading. We will measure the dielectric constant of these composites and compare those results with those from refined computational models to determine the dielectric constant of individual BTO nanoparticles.

Advisor(s): Albert Dato and Peter N. Saeta.

Team: Charles Burke Dawson ’19, Alejandro E Baptista ’18, Andrew Mather Bishop ’18, Benjamin I Lehman ’18, Richard Arthur Liu ’18, and Lupe Maria MacIntosh ’18.

The team designed an algorithm that loads electrons onto silicon heterostructure quantum dots to initialize a qubit. The algorithm relies on an internal electrostatics model of the dot system to navigate through the space of voltages controlling the potential energy landscape. It also uses a modified second model to simulate real experimental feedback during development and testing, since the team did not have access to the laboratory setup. The algorithm queries the internal model to determine a path through voltage space that results in a specific charge configuration for the quantum dot system.

Advisor(s): Gregory A. Lyzenga.

Team: Brynn Elise Arborico ’17, Amy Frances Brown ’17, Max James Byers ’17, and Kathleen Elizabeth Kohl ’17.

Raman spectroscopy can discriminate between healthy and cancerous breast tissue. This report examines the potential to implement an inexpensive commercial Raman system and hand held probe for the in vivo, real-time detection of cancer in surgical margins. First, 785 nm and 1064 nm systems are compared, and the 785 nm system is identified as preferable because of its greater resolution, superior signal-to-noise ratio, and extended spectral range. A method to eliminate the greater fluorescent contribution resulting from excitation with 785 nm is described. In a second experiment, Raman spectra are collected along lines that transect cancer margins of two patients following lumpectomy. Spectra are classified as either healthy or cancerous according to histological assessment of collection location, and 15 spectral bands are identified as the most descriptive of spectral variation. Discriminant analysis performed on the two primary principal components of the bands is shown to classify tissue as healthy or cancerous with 100% accuracy. A third experiment expands the classification to naive spectra obtained from tissue samples of additional patients to assess cross-patient classification. Finally, a fourth experiment achieves 100% classification using substantially reduced spectral bins, demonstrating the potential for real-time in-vivo cancer detection.

Advisor(s): Michael C. Storrie-Lombardi.

Team: Sarah Marie Anderson ’17, Alexander Felipe Echevarria ’17, Nathaniel Loren Miller ’17, Connor Emerson Stashko ’17, and Willie Correa Zuniga ’17.

Barium titanate, or BTO, is widely used as a dielectric material due to its high dielectric constant, which typically ranges from 1500 to 2000 in bulk. Despite its prevalence as a dielectric material, current literature remains unclear on how BTO nanoparticle size impacts the dielectric constant, particularly for non-sintered, discrete nanoparticles. Other research groups have reported BTO nanoparticle dielectric constants ranging from 135 to 5000, with no indication of the uncertainty. Our team previously attempted to measure the dielectric constant of BTO nanoparticles by loading them in an epoxy composite, measuring the effective dielectric constant of the mixture, and using COMSOL modeling to extract the dielectric constant of the nanoparticles themselves. Experimental observations and additional COMSOL modeling indicated that agglomeration of BTO nanoparticles within the composites raised the effective dielectric constant such that determination of the BTO nanoparticle dielectric constant was impossible using our existing methods.

In order to overcome the agglomeration issue and allow for the extraction of the BTO nanoparticle dielectric constant, we developed a ball-milling procedure to eliminate BTO nanoparticle agglomerates before loading them into the epoxy. We examined two volume fractions of BTO that can be achieved with this procedure, and were  able to determine the dielectric constant of 500 nm BTO nanoparticles in a 20% volume loading composite to be 225 ± 75 for our nanoparticles, assuming 0% agglomeration. Finally, we also developed more robust COMSOL models for understanding the effects of BTO agglomerate size on the effective dielectric constant.

Advisor(s): Albert Dato and Richard Campbell Haskell.

Team: Isabel Ann Martos-Repath ’18, Marisol Nora Beck ’17, Jonas Leif Kaufman ’17, Cesar Jose Orellana ’17, Carmel Jia Zhao ’17, and Robin Bendiak ’17.