MICRO Abstracts

Would you like to learn more about the sort of research MICRO students work on? The abstracts below describe the work MICRO students performed in the Fall of '21.


Anastacia De Gorostiza

In the past two decades, the development of silicon photonics has revolutionized the field of integrated circuits by using silicon to guide light. Furthermore, silicon-on-insulator provides a low-cost and high-efficiency material platform for use in photonic integrated circuits (PICs). However, a common problem in PICs is the size difference between a waveguide and the light source, typically a laser or LED coupled to the chip through an optical fiber. Thus, a main issue often being studied in the field today is how to optimize the coupling of light emitted from a laser or LED to these significantly smaller waveguide structures. In order to achieve higher coupling efficiency between these devices in a PIC, researchers have adopted two main techniques: edge coupling and grating coupling. For the past semester, I have been working on increasing the coupling efficiency from LED sources into PIC chips designed by the Electronic Materials Research Group (EMAT) at MIT.


Eyobel Haile

Metal Additive Manufacturing by selective laser melting (SLM) is a promising technology for large scale manufacturing, but is also a time consuming, expensive, and labor intensive process. From this perspective, it is important to make sure that the printing process goes as smoothly as possible and produces consistent parts to avoid additional processing steps or failed prints. In this project, in-situ quality control techniques were implemented in order to inspect the structural integrity of printed parts during the process. The hardness data of certain materials of interest was collected and used for determining the local quality of printed parts while controlling for all other variables. Linking measured hardness to temperature gradient, porosity and part defects enable identification of localized defects in the print. Such a method opens new opportunities for reducing the occurrence of failed prints and improve printing quality.


Joshua Chaj Ulloa

Deep Neural Networks Hyperparameters Optimization within Photonic Meta Surface Devices Simulations

Joshua Chaj Ulloa, Sensong An, JueJun Hu

Photonic systems and devices are applicable to various biomedical research applications, with the overarching scope of the field being the study of light and its applications to biosensing, metamaterial optimization, and optical engineering. However, for photonic devices to be applicable to their specific field applications, customization and specification with regards to their device geometry and optical response must be studied and troubleshot for their certain optical properties to be tailor-made. Therefore, researchers have applied the field of neural network machine learning algorithms toward the specific development and optimization of various physical, spectral, and performance characteristics of photonic devices to enhance these properties. Our project focuses on the development of neural networks that will allow us to recognize hidden patterns and correlations for the photonic meta surface devices experimental datasets. Where an overall exploration will be conducted of the effects specific hyperparameter tuning will have on the neural network functionality and performance as this tuning allows for the specific optimization of the hidden layer and overall structure of the deep neural network. The analysis will apply specific hyperparameter tuning methods such as Grid Search, Random Search, and Bayesian Optimization toward monitoring the performance of the complex neural networks developed. This unique approach provides promise toward the enhanced simulation and design development process of photonic meta surface devices toward their specific biomedical research applications.


Nicholas Layman

Using Dynamic Time Warping for Identification of Peptides in Affinity Selection Mass Spectrometry

Nicholas Layman, Somesh Mohapatra, Rafael Gómez-Bombarelli

One of the early stages of drug discovery is the identification of a protein's specific binders to aid in localizing medicine absorption and to ensure effective treatment. Identifying specific binders one at a time is straightforward but extremely time-consuming so we use affinity selection - mass spectrometry (AS-MS) to test many potential binders at once and obtain a collection of suitable binders. To make this process more robust we repeat it a few times per protein and combine these trials into one average dataset. These datasets cannot be combined naively due to small temporal variations in a filtering stage of AS-MS. To overcome these challenges, I implemented and optimized dynamic time warping methods to obtain and analyze a typical dataset. The fully functional and accessible code base I developed enables process automatization as the first step towards systematic binder selection from any AS-MS dataset.


Rachel Myers

Determining the genomic drivers of polystyrene nanoparticle interactions in cancer cells

Rachel Myers1, Dr. Joelle P. Straehla2,3,4,5,6, Dr. Paula T. Hammond2,3

1 Department of Chemical, Biochemical and Environmental Engineering, University of Maryland Baltimore County; Baltimore, MD
2 Koch Institute for Integrative Cancer Research, Massachusetts Institute of Technology; Cambridge, MA
3Department of Chemical Engineering, Massachusetts Institute of Technology; Cambridge, MA
4 Broad Institute of MIT and Harvard; Cambridge, MA, USA
5 Department of Pediatric Oncology, Dana-Farber Cancer Institute; Boston, MA
6 Division of Pediatric Hematology/Oncology, Boston Children’s Hospital; Boston, MA

Cancer nanomedicine is a promising venture for the delivery of therapeutics to cancerous tissue, as nanoparticles (NP) have been shown to increase circulation time, reduce early degradation and clearance, improve specificity, and lower off-target effects of the drug. However, the high variability between cancer cell lines presents an issue with understanding which NP surface chemistries result in successful nanoparticle trafficking, penetration, and payload delivery. This project focuses on using machine learning algorithms to determine the genomic drivers of NP-cancer cell interactions observed in a massively parallel pooled screening. Here, we explore techniques to translate the high-throughput cell screening data into informative genomic nanoparticle trafficking networks. The use of random forest algorithms, clustering, and protein-protein interaction databases to identify biomarkers for interactions in polystyrene nanoparticles is studied. This novel paradigm to study NP-cell interactions may provide an efficient, genomics-based approach to design nanocarriers for cancer treatment.


Samuel Figueroa

Refractory Metal Nanocrystalline Alloy Development using Nanophase Separation Sintering

Samuel David Figueroa1, Christian Edward Oliver2, Yannick Naunheim2, Christopher Schuh2
1University of California, San Diego, La Jolla, CA 92093, United States
2Department of Materials Science and Engineering, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA 02139, USA

Nanocrystalline alloys are of great interest due to their exceptional physical properties. Recently, a novel nanoscale phase separation sintering (NPSS) technique has shown production of nanocrystalline binary alloys. NPSS studies have demonstrated that nanocrystalline powder particles supersaturated with a secondary element begin to phase separate upon heating to form solid-state necking to promote rapid diffusion of atoms. This project aims to expand the applicability of NPSS to the molybdenum-chromium (Mo-Cr) binary alloy system using a master sintering curve model to predict densification behavior. This work is based on preliminary experimental characterization of the Mo-Cr NPSS system. Attempts to implement a master sintering curve analysis imply this method helps determine average kinetics, but it is not useful for modeling the various mechanisms of NPSS. Future work will focus on the use of more sophisticated analytical models in tandem with experimental investigation to characterize nanocrystalline structures in the Mo-Cr sintering system.


Tobi Majekodunmi

Finite Element Analysis of a Three-Dimensional Solid-State Battery

Tobi Majekodunmi1*, Richard Church2, Professor A. John Hart3

1Department of Mechanical Engineering, University of Maryland, Baltimore County, Baltimore, MD
2Department of Materials Science and Engineering, Massachusetts Institute of Technology, Cambridge, MA
3Department of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, MA

Lithium-ion batteries are ubiquitous within today’s energy storage applications, particularly in the emerging field of electric vehicles. To be competitive with existing transportation technologies and overcome the “range anxiety” felt by many consumers, electric vehicles require batteries with higher energy densities than what is currently available. Without exploring other chemistries, the simple solution is to utilize thick electrodes. However, increasing electrode thickness in planar cell designs comes at the expense of power density, which negatively impacts the car’s acceleration. MIT’s Mechanosynthesis Group is developing a solid-state battery for Lamborghini’s pilot electric supercar. This battery utilizes a 3D, interdigitated geometry and highly conductive carbon nanotubes to simultaneously increase energy density and power output. This project performed discharge simulations with COMSOL modeling software to guide experimental design and fabrication by analyzing how the battery’s geometry, ionic conductivity, and electronic conductivity impact performance.