2022 REU Symposium

Jenny Hoffman
Mentor: Jason Hoffman

Abstract: Vanadium dioxide (VO₂) undergoes an insulator-to-metal transition (IMT) at around 340 K, which makes it an interesting material for applications in "smart windows," low-power computing, and night-vision goggles. When VO₂ thin films are deposited on substrates such as sapphire (Al₂O₃) or titanium dioxide (TiO₂), the substrate-induced strain changes the VO₂ properties, including the crystalline arrangement and the electronic transport behavior. The goal of this project is to use molecular beam epitaxy to deposit VO₂ thin films on the c-, m-, a-, and r-planes of Al2O3 and (001) TiO₂. After growth, we use x-ray diffraction to determine the film orientation and polymorph. We demonstrate the epitaxial stabilization of VO₂(B) films on r-plane sapphire.

Tionna Tapaha, Navajo Technical University (Crownpoint, NM), Rohan Thakur, Harvard University (Cambridge, MA), Robinson Tom, Harvard University (Cambridge, MA), and Dr. David Weitz, Harvard University (Cambridge, MA)
Mentor: David Weitz

Research Focus: Microfluidics
Highlight and Abstract: Use of microfluidic drops as self-contained miniature "test tubes" for biological experiments, allows for greatly increased throughput while reducing reagent volumes compared to traditional methods. Traditional droplet sorting uses dielectrophoresis and uses different systems to promote selective droplet sorting, which is typically slow or lower throughput. In this project, we are developing a new method for sorting droplets by first selectively charging the droplets and then sorting them. This means that we can increase droplet sorting throughput by scaling up the size of the sorting region. To determine if our electrophoretic method is working, we first check if the drops are charged and then if the droplets deflect in a uniform field. To enable charging of the droplets, we picoinject drops with an electrical signal DC offset. By charging half of the droplets, all of the droplets, and then none of the droplets, we were able to test if the device was effectively charging. If droplets were properly charged, they produced a thin filament instability between them and the uncharged drops, allowing the drops to coalesce. Once evidence of charging was obtained, we tested the droplets for deflection, in order to determine that they could also be selectively sorted.

Zabari Bell, Rohan Thakur, Robinson Tom, and Dr. David Weitz, Harvard University (Cambridge, MA)
Mentors: Robinson Tom, Rohan Thakur

Highlight and Abstract: Many biological samples are heterogeneous. By individually encapsulating each cell in a droplet using droplet microfluidics, we can perform single cell sequencing to uncover the genomic and transcriptomic differences between individual cells. However, sequencing depth is shared for each cell processed. For many rare cell applications, this means that single cell sequencing alone will not provide sufficient information for the rare cell of interest. Therefore, droplet sorting is a helpful technology because only the cell of interest is being sequenced. This enables high resolution of the cell of interest. In this project, we are developing a new method of sorting droplets by charging the droplets selectively and then sorting them. Our experimental approach uses electrophoresis which should have a higher throughput than dielectrophoresis.

To determine if our electrophoresis method works we validate if the droplets are being charged and if they deflect in a uniform electric field. To charge droplets we picoinject droplets with an electrical signal with a DC offset. This inparts a charge into the droplets we inject half of the droplets with this wave form and view the entire collection of the droplets. When a charged droplet is adjacent to an uncharged droplet, the uncharged droplet becomes polarized and a thin film instability forms leading to droplet coalescence. When we look at a collection of droplets where half were charged, we see significant coalescence, implying that charging was successful.

Margaret Lashutka, Aric Lu, Jennifer Lewis
Mentor: Jennifer Lewis

Highlight and Abstract: There is currently an organ shortage in the United States, leaving many who are in need of organ transplants without any organs for transplantation. One potential solution to this issue is creating artificial organs, which can be done by 3D bioprinting genetically engineered stem-cell derived tissues. However, a challenge that comes with this is the difficulty of building genetically heterogeneous tissues that contain multiple cell types with distinct patterns of gene expression. This is a feat that is necessary for building certain organs such as the heart, which contains ventricular heart muscle comprised of primarily contractile cardiomyocytes but also conducting cardiomyocytes. Therefore, we have decided to address this challenge by coupling 3D bioprinting with electroporation, a genetic engineering technology that involves applying an electric field to these cells to form temporary pores inside of the cell membrane. This then allows us to genetically engineer cells as they are printed by inserting DNA into the cells through these pores. By selectively modifying these cells during the printing process, we are able to create the patterns of gene expression that we need to build complex tissues by using 3D bioprinting. This summer, we worked to develop a nozzle that would pass cells and DNA through two electrodes, thus forcing the expression of specific genes.

Mentor: David Weitz

Highlight and Abstract: There is currently an organ shortage in the United States, leaving many who are in need of organ transplants without any organs for transplantation. One potential solution to this issue is creating artificial organs, which can be done by 3D bioprinting genetically engineered stem-cell derived tissues. However, a challenge that comes with this is the difficulty of building genetically heterogeneous tissues that contain multiple cell types with distinct patterns of gene expression. This is a feat that is necessary for building certain organs such as the heart, which contains ventricular heart muscle comprised of primarily contractile cardiomyocytes but also conducting cardiomyocytes. Therefore, we have decided to address this challenge by coupling 3D bioprinting with electroporation, a genetic engineering technology that involves applying an electric field to these cells to form temporary pores inside of the cell membrane. This then allows us to genetically engineer cells as they are printed by inserting DNA into the cells through these pores. By selectively modifying these cells during the printing process, we are able to create the patterns of gene expression that we need to build complex tissues by using 3D bioprinting. This summer, we worked to develop a nozzle that would pass cells and DNA through two electrodes, thus forcing the expression of specific genes.

Samantha Francis, Justin Platero, Dylan Barber, Jennifer Lewis, Thiagarajan Soundappan
Mentor: Jennifer Lewis

Highlight and Abstract: Climate change has motivated a rapid advance in sustainable and carbon-neutral energy sources that avoid the environmental issues associated with fossil fuels. Leading candidates include solar and wind for power, which is more environmentally friendly than previous techniques. However, solar and wind energy are intermittent energy sources, meaning that they do not provide a constant source. A reliable battery storage is needed to convert to a majority wind- and solar-based electrical grid. Redox flow batteries are a promising candidate for renewable energy storage since they are cost effective and scalable. They rely on conversion of electrical energy to chemical potential energy in an electrolyte solution for ready conversion back into electrical energy. In this work, we prepared a library of 3D-printed lattices of varying geometry for use as electrodes in redox flow batteries. Two lattice geometries were designed with 200 and 300 μm pitch in the flow direction to investigate the role of filament spacing in determining battery efficiency. Then, lattices were subjected to post-processing to make them conductive for battery usage. We anticipate that these novel lattice designs will prove to influence the efficiency of flow through the lattice and improve battery performance.

Justin Platero, Samantha Francis, Dylan Barber, Jennifer Lewis, Thiagarajan Soundappan
Mentor: Jennifer Lewis

Highlight and Abstract: The drive for renewable resources is at an all-time high because of increasing greenhouse gas emissions which causes climate change. We have developed methods of harnessing renewable energy such as solar and windmills. While a promising alternative to fossil fuels, these methods provide power intermittently. To counter this, advanced energy storage technologies must compensate at times when wind and sunlight cannot meet our grid-scale demands. Flow batteries, consisting of two porous electrodes and a scalable electrolyte tank, are an emerging technology for large scale energy storage. A key scientific and engineering opportunity in flow battery design is the relationship between porous electrode geometry and battery efficiency. To better understand this relationship, we 3D print battery electrodes in the Lewis lab. We use a graphene nanocomposite ink to create lattice structures with different geometries for use as flow battery electrodes. We have been investigating the rheological properties of the inks by studying how graphene content impacts yield stress and shear-thinning behavior. By measuring the storage and loss modulus of our ink through a rotational oscillation test we found that the amount of solvent and graphene content affects ink composition. We also measured the viscosity of our ink using continuous rotational test, revealing how viscosity is affected by amount of solvent and graphene content. The rheology data we've gathered will allow us to optimize our method for making ink and improve how well it extrudes during printing. This will lead to new complex lattice structures for our 3D printed electrodes.

Rothiel Davis II, Bobby Haney, Thomas Cochard, David Weitz
Mentor: David Weitz

Highlight and Abstract: With the use of a 2D transparent artificial porous media, it was able to visualize and quantify a new type of transport of residual oil in the case of surfactant flooding. By decreasing the interfacial tension between the oil phase and the flooding fluid by twenty times compared to the water flooding base case it was observed that an initially trapped oil emulsion of multiple times the pore size starts to be eroded at a critical capillary number. Images were taken to capture the erosion and formation of oil droplets that varied in size based on pore geometry which demonstrated its impacts on the overall oil transport leading to a significant improvement in the oil sweeping efficiency. Specifically, in the report, an identical capillary number of different dynamics indicated that decreasing the interfacial tension caused a change in the mechanical properties of the oil and flooding fluid interface leading to an increase in deformability of the oil phase. As a result, the oil emulsion can deform through the pore network, and be sheared out into free drops smaller than the pore size. Those free drops are considerably less subject to capillary forces and move freely in the pore network at a velocity close to the flooding fluid velocity. The oil droplet formation leads to an increase in recovered oil fraction but also speeds up drastically the oil transport. This experimental evidence of the spontaneous oil droplet formation has implications for a range of oil phase flow processes in natural and engineered porous media.

Hunter Shepard, Will Wang, Frans A. Spaepen
Mentors: Franz Spaepen, David Weitz

Highlight and Abstract: The purpose of this project is to improve crystal identification and identify structures of experimental data in colloidal science. Colloidal processes are essential to many different industries, including the food, pharmaceutical, agrochemical, cosmetics, and polymer sectors, and they serve as the foundation for a variety of goods. Industrial crystallization methods are very interested in the nucleation of crystals from solutions, particularly where the form of the crystals is crucial to the applicability. However in experimental data, there are thermo-fluctuations, and noise is a simulation or error or interruptions that naturally happen. Through coding in python and constructing three different structures and using three different methods, by adding noise, develops different datasets and visualizations which are used to display and examine output data from simulations using particles This system is placed via modeling and computational analysis on a lattice. The findings may be contrasted to experimental findings and might provide greater insight into the crucial process variables needed to design a certain form in a given crystallization method.

Hunter Shepard, Will Wang, Frans A. Spaepen
Mentors: Franz Spaepen, David Weitz

Highlight and Abstract: To further our understanding of the characteristics of biomolecular condensates, we are specifically using PolyA, a mRNA strand with a tail of several adenosine monophosphates, and analyzing their interaction with PEG-coated particles, PEG being a crowding agent. It is often observed that the PEG-coated particles stay on the surface of the condensate rather than within the condensate. With the intention to eventually do microrheology experiments, there is evidence to believe that the relative size of the particle surface coating molecules plays an important role in the affinity between the condensate and the particle. Using a PolyA-PEG system allows for easy control of the size of the coating molecules and other components; subsequently, finding the ideal circumstances to where analysis by microrheology is maximized is the goal. Analyzing the fraction of each probe particle that is embedded in each respective condensate will give us insight into how each molecule size differs in affinity between the condensate and particle.

Erin McGee, C.J. Xin, Marko Lončar
Mentors: Franz Spaepen, David Weitz

Highlight and Abstract: Thin film lithium niobate has been a material of interest in the optics and photonics community for the last several decades, primarily due to its strong nonlinearity, high refractive index, and electro-optical properties, allowing it to be used in high speed telecommunications as well as be integrated into quantum technologies. In particular, nanofabricated waveguides etched in LN can allow for the conversion of frequencies of light, enabling shorter wavelengths, and therefore less loss in communications. These waveguides, however, are extremely sensitive to changes in geometry and treatments during the nanofabrication process. The goal of the experiments that we are carrying out is to find any correlations between the etching and cleaning conditions and the measured geometry and refractive indices of the lithium niobate ridge waveguides. In our initial findings, we established that the relationship between annealed and unannealed thin film lithium niobate samples and their refractive indices is consistent with earlier data and that certain treatments to the silicon layer can change the refractive index by as much as 0.02. In addition to this, we found that the best practices for characterization using high aspect ratio tips with Atomic Force Microscopy include scanning perpendicular to the waveguide to reduce the asymmetry and using the relative angles of the waveguide to determine the sharpness of the tip.

Marko Lončar

Mentor: Eric Mazur

Zofia Adamska, Megan C. Engel, Michael P. Brenner
Mentor: Michael Brenner

Highlight and Abstract: Understanding and being able to control complex non-equilibrium systems is crucial for many applications in biology and nanotechnology. One particular class of out-of-equilibrium problems involves a protocol Λ(t) that drives a system between its initial and final states. The goal is to find a protocol that produces the desired distribution of thermodynamic properties. Using JAX-MD¹, a python molecular dynamics library with built-in automatic differentiation and just-in-time compilation, we have developed a method to find optimal protocols for single-molecule pulling experiments that minimize the dissipative work and protocols that minimize the error in free-energy estimation based on the Jarzynski equality. Our method produces results that agree with previous findings. Compared with the previous approaches, our method can explore more complex biomolecules driven farther from equilibrium. We used this new method to develop an iterative algorithm that uses JAX-MD simulations along with experiments to improve the accuracy of free energy landscapes reconstructions from single-molecule pulling experiments.

Zofia Adamska, Megan C. Engel, Michael P. Brenner
Mentor: Michael P. Brenner

Highlight and Abstract: Understanding and being able to control complex non-equilibrium systems is crucial for many applications in biology and nanotechnology. One particular class of out-of-equilibrium problems involves a protocol Λ(t) that drives a system between its initial and final states. The goal is to find a protocol that produces the desired distribution of thermodynamic properties. Using JAX-MD¹, a python molecular dynamics library with built-in automatic differentiation and just-in-time compilation, we have developed a method to find optimal protocols for single-molecule pulling experiments that minimize the dissipative work and protocols that minimize the error in free-energy estimation based on the Jarzynski equality. Our method produces results that agree with previous findings. Compared with the previous approaches, our method can explore more complex biomolecules driven farther from equilibrium. We used this new method to develop an iterative algorithm that uses JAX-MD simulations along with experiments to improve the accuracy of free energy landscapes reconstructions from single-molecule pulling experiments.

Katelyn Wilson, Layla James, Jean Serrano Flores, Yang Wang, David Weitz
Mentor: David Weitz

Highlight and Abstract: We have started research on developing a microfluidic-based system for droplet, digital polymerase chain reaction (PCR) as a method to detect pathogenic RNA in clinical blood samples. Based on this, we are implementing digital droplet PCR (ddPCR) as a novel assay for ultra-sensitive COVID diagnosis. We also incorporate gold nanorods to perform nanoplasmonic-based heating, while protecting droplets with PCR reagents with a double-emulsion system.

Traditional, double emulsions are generated with glass-capillary devices which lack reproducibility and scalability. To solve this, we focus on translating the double emulsion generation into a PDMS-based device. However, to achieve adequate spatial wetting (hydrophobic and hydrophilic regions) required for double emulsions, we implement hydrophobic PDMS devices and plasma oxidation. As such, we seal the device and allow plasma to diffuse into the outlet of the device to solely render the downstream part of the channels hydrophilic. The implementation of microfluidic devices will be fundamental to this project since it provides the capabilities of amplifying and identifying bacteria in blood samples. By doing so, it generates droplets of the samples to then perform digital PCR and on-chip sorting.

Desvaun Drummond, Joshua Sanchez, Riccardo Comin

Abstract: The project goal is to study the interaction of magnetism and structure of 2D magnetic materials, which provides a new opportunity for nanoelectronics. The central material under investigation is nickel iodine, which has a helical magnetic structure that is tightly coupled to the lattice. We will use chemical vapor deposition to grow single atomic layer crystals, starting from simple elemental powders and using high temperature furnaces. Novel micro scale strain devicesfrom etching silicon wafers will be developed using the MIT.nano fabrication facility. The samples will be attached to strain devices to tune the magnetic properties by stretching the sample. The stretched sample will be investigated with optical characterization techniques using polarized and laser light, including MOKE and Raman scattering, to measure magnetic properties and structural tuning. Through this unique combination of strain tuning and optical characterization, we gain insight into how information can be stored in single atomic layer materials. To adapt our cryostat to make these strain measurements, we will also build a new electronics interface.

Katarina Cheng, Vic Feng, Minlan Yu
Mentor: Minlan Yu

Highlight and Abstract: P4-programmable switches have been gaining popularity as a network accelerators for a variety of applications including database queries, network load balancing, and even machine learning. They offer the high performance of a switch with more flexibility due to the programmability, but the programmers are heavily constrained by the switch hardware’s small memory and limited operations, which makes building complex systems difficult.

In this work, we create a data structure library that makes building these systems easier for P4-programmers, beginning with a ring buffer. We implemented an API for this ring buffer using P4 and Python both in the data plane and the control plane, with a pull-based approach. Further, for the data plane program, we implemented a shared queue when multiple ring buffers are utilized to minimize the number of pipeline stages used. We observed that beyond 4 buffers, the non-shared buffers use one more stage for every 4 new buffers, while the shared buffer sees no increase in stages, as a result of each stage allowing a maximum of 4 registers. There was no significant difference in SRAM usage between the two methods, as it is determined by the total capacity of the buffers. This ring buffer implementation will allow programmers to implement systems that require ring buffers, such as for request and packet queueing.

Andrew Huang, Vic Feng, Danny Chen, Minlan Yu
Mentor: Minlan Yu

Highlight and Abstract: Hardware accelerators have been shown to be a promising area to find performance gains in as Moore's law comes to an end. However, because these technologies trade off flexibility for performance, they have not yet become widely used in industry. We propose that the financial industry is a promising area for growth of hardware accelerators. In the financial area, there is a "race to zero" latency, because the companies that are able to achieve the lowest latency are able to profit on fleeting arbitrage opportunities faster than their competitors. This means that they may be promising first adopters of the cutting edge of hardware acceleration. Our project seeks to make a hardware accelerator for finance applications in a programmable switch. We have implemented an orderbook with p4 on an Intel Tofino that has a latency of about 100 nanoseconds. Tofino's hardware constraints make it so that traditional orderbook implementations cannot fit into the switch. Our idea is to have the switch work together with a CPU. The switch will act as a "fast path" for the orderbook, while the CPU is responsible for maintaining correctness when the switch fails. Latency measures of the orderbook implemented on a switch are not faster than the state of the art orderbook implemented on an FPGA (44 ns). However, we are currently looking at applications like pre-trade risk check, where we can take advantage of the larger throughput in a switch as compared to an FPGA. Overall, more work in this area exploring how different ASICs can be used as a hardware accelerator for finance applications would allow us to better understand how to incrementally transition towards domain specific hardware.

Rea Tresa, Joshua M. Brockman, Nicholas Jeffreys, David J. Mooney
Mentor: David Mooney

Highlight and Abstract: Immunotherapies have exhibited significant promise in the treatment of human cancer; however, most patients fail to respond. Recent reports suggest that patients who do respond to immunotherapies often form tumour-associated tertiary lymphoid structures (TLS). TLS are lymph-node like structures that form at sites of chronic inflammation, including tumors, and are associated with a positive prognosis in many human cancers. Herein, we utilise chemokine-infused collagen-alginate cryogels as an engineered biomaterial platform to induce TLS formation on demand. The specific goal of this summer research was to develop a microscopy-based method to characterise the timing and dose-response of biomaterial-induced T cell recruitment and TLS formation.

Cryogels infused with different doses of a chemokine cocktail were injected into the subcutaneous flank of mice. These gels were extracted between 1 and 6 weeks post implantation and subjected to immunofluorescence analysis. We observe lymphocyte recruitment and TLS formation in the vicinity of the gel as early as one week after injection. Preliminary results show a decrease of T cells per unit area with time, and a biphasic dose-response in T cell recruitment in response to increasing chemokine concentration. This work provides quantitative understanding of lymphocyte recruitment and TLS formation in response to the biomaterial system. Future work will focus on elucidating the mechanisms behind TLS formation and harnessing TLS formation to improve responses to existing immunotherapies.

Mikayla Jackson, Einat Vitner, David J. Mooney
Mentor: David Mooney

Highlight and Abstract: An inflammatory response is a protective strategy for the host in response to detrimental attacks such as microbial infection, tissue injury, and other noxious conditions. Chronic inflammation occurs when the immune system is over-stimulated constantly, and significant damages that are done to the host are mediated by the person’s inflammatory response itself, and not by foreign invaders. The therapeutic goal of our lab is to suppress this response, and regulatory T cells are utilized as a key player in immune suppression. Cryogels are being used as a biomaterial approach, and these gels have desirable physical and biocompatible properties. Advantages include the sustained release of compounds from the gels that will spare the need for multiple injections, enrichment of the target cells in the scaffold that will decrease the off-target effect, and the ability for in vivo expansion of regulatory T cells. Our lab aims to establish a regulatory T cell induced in vitro model, engineering of a cryogel-based regulatory T cell induced system, and delineating the effect of recruitment factors on regulatory T cell expansion. As a result of our experiments we have established a simple and fast method to detect regulatory T cells in vitro that will be utilized in future experiments to determine the percentage of regulatory T cells. In addition, we’ve found that anti-CD3 and anti-CD28 antibodies, while conjugated to our cryogels, can be used to induce T regulatory cells. Lastly, CXCL9/10/11 have been identified as recruitment factors that don’t interfere with T regulatory cell induction. The next step is to implement this system in vivo.

Leonardo Claure, Li Qiang, Jia Liu
Mentor: Jia Liu

Highlight and Abstract: Brain organoids are three-dimensional (3D) models derived from human induced pluripotent stem cells (hiPSCs), which are employed to further our understanding of human brain development and disease progression. It is crucial to understand the evolution of both electrophysiological function and gene expression over time, and how it differs from the in vivo counterparts. Previously, microelectrode array (MEA) plates and calcium imaging methods have been used to acquire electrical activities from the active neuronal cells that compose brain organoids. However, the most common issue encountered with these methods is their lack of high resolution and long-term stable readings, which precludes our ability to fully understand the function of the organoids during development. To address these challenges, we have created a flexible mesh-like nanoelectronics system that can be integrated into the growing brain organoids to create a cyborg organoid system. The flexible nanoelectronics can withstand the forces associated with organogenesis and it bends to adjust, which is what allows for accurate real time readings over extended periods. Le Floch et al 2022 demonstrated that the integration of these devices does not affect brain organoid development by showing that gene expression and the cell types present remain the same between cyborg brain organoids and control brain organoids. When combining RNA sequencing and the nanoelectronics we acquire information about which genes are being expressed and which cells are electrically active over time. We are interested in understanding the impact of the HOPX gene in the development of brain organoids and their electrical function. Previous research in non-human gyrencephalic, brains that have cortical folds, animals has shown that the HOPX gene plays a key role in the formation of gyri and sulci. Matsumoto et al. 2020 showed that in ferrets HOPX+ cells preferentially accumulate around regions that will form gyri whereas HOPX- cells do not. We wanted to see if this correlation between the HOPX gene and cortical folding remained the same in the human cyborg organoid system and how it differs from the ferrets. We started with the generation of HOPX gene positive brain organoids and HOPX gene negative brain organoids from two different iPSC cell lines, which we cultured until we could introduce the flexible nanoelectronic. Then the iPSCs undergo neural induction so that the cells would adopt the cortical neuron cell fate and we continue to expand the culture as we differentiate the cells to their neuronal fate. At this time the cells could reorganize to form a 3D structure of differentiating neurons with the electronics embedded within the cells. Once neurons are fully differentiated, we can collect electrophysiology data from the cells so any time a cell fires an action potential this is detected by the nanoelectronics and is represented in the form of a spike detected by a sensor at a specific time point. In combination with single cell RNA sequencing, we can determine which cells are active and what genes are being expressed at specific time points along development. By establishing the relationship between electrophysiology and gene expression in brain organoids that allows us to understand the impact of manipulations to the system. Ultimately by furthering our understanding of brain organoids we hope to shift away from animal models which fail to fully capture the genetic, molecular, structural, and functional complexity of the human brain.

Works Referenced:
Le Floch, P., Q. Li, Z. Lin, S. Zhao, R. Liu, K. Tasnim, H. Jiang, and J. Liu, "Stretchable mesh nanoelectronics for three-dimensional single-cell chronic electrophysiology from developing brain organoids," Advanced Materials 34(11), 2106829 (2022)

Matsumoto, N., S. Tanaka, T. Horiike, Y. Shinmyo, and H. Kawasaki, "A discrete subtype of neural progenitor crucial for cortical folding in the gyrencephalic mammalian brain," ELife, 9, 1–26 (2020)

Velin Kojouharov, Alyssa Hernandez, Perrin Schiebel, Robert Wood
Mentor: Robert Wood

Highlight and Abstract: Many insects have the ability to traverse rough, inclined surfaces, yet legged robots still fail to match their performance. The ability to traverse these surfaces is particularly important in insect-scale robots that navigate outdoor landscapes and inspect confined spaces. This work introduces a new redesigned 6-legged version of the Harvard Ambulatory Microbot (HAMR). Preliminary tests indicate that the 6-legged platform outperforms the quadruped robot in terms of stability on inclined surfaces. Furthermore, this new robot can shed light on the role of aspect ratio in locomotion over rough and inclined surfaces, inspired by the different body shapes of beetles. Using an open loop control strategy, the length and width of the robot can be adjusted to find an optimal aspect ratio for different terrains by keeping the area of the rectangle formed by the outer leg contact points constant. This lightweight, stable redesign of the HAMR platform can be used to improve the locomotion of insect-scale legged robots while also providing insight into the role of aspect ratio in beetle locomotion.

Lauryn Whiteside, Michael Bell, Robert Wood
Mentor: Robert Wood

Highlight and Abstract: Biofilm or biofouling is the accumulation of microorganisms, plants, algae, or small animals in unintended places such as on the hulls of ships. When a ship hull becomes overly bio-fouled this has negative impacts on the ships fuel usage increasing the CO₂ production and slowing the ships movement. A better understanding of the forces and materials of wipers and brushes used to remove biofouling for this kind of robotic system is needed to ensure that effective cleaning can be performed without damaging a ships’ coatings or hull. In order to study these parameters a three axis gantry was used to test the effectiveness of different wipers and brushes at various forces. This testing was done on sample biofouled steel plates grown in the lab. Fabrication of simulated surfaces that replicated force required to remove biofilm from plates were created to enable more frequent testing. A tank robot with a magnetic tread was designed using 3D printing of flexible filaments and injection molding. Forces required to remove various levels of biofouling were determined for three wipers and two brushes. Several materials were identified to accurately replicate a biofouled surface. Designing robotic solutions that allow for underwater cleaning of biofilm on ship hulls while they are underway prevents the introduction of non-native and invasive species found in biofilm to ship port marine environments. The ability to continuously remove biofouling while the ship is underway will mitigate the effects of biofouling on CO₂ production and fuel utilization.

Mohammed Sbai, Yang Yi, Katia Bertoldi
Mentor: Katia Bertoldi

Highlight and Abstract: When making soft robots that function mainly through deformation it is important to have a precise shell thickness in order to control the movements of the robot. There are multiple methods to create those shells ( Rotational, injection, and blow molding) but most of them are expensive and mainly designed for industry use and not for laboratory use where it is important to have precise measures.

We are trying to experiment with different coating methods to know which one gives us the best and most precise thin shells ( Micro level) for soft robotics use. The resulted data were used to create a soft rotor that is made completely from soft materials.

Abigail R. Lockhart-Calpito, Lucas F. Gerez, Harrison Young, Conor J. Walsh
Mentor: Conor Walsh

Highlight and Abstract: When making soft robots that function mainly through deformation it is important to have a precise shell thickness in order to control the movements of the robot. There are multiple methods to create those shells ( Rotational, injection, and blow molding) but most of them are expensive and mainly designed for industry use and not for laboratory use where it is important to have precise measures.

We are trying to experiment with different coating methods to know which one gives us the best and most precise thin shells ( Micro level) for soft robotics use. The resulted data were used to create a soft rotor that is made completely from soft materials.

Abigail R. Lockhart-Calpito, Lucas F. Gerez, Harrison Young, Conor J. Walsh
Mentor: Conor Walsh

Baran Mensah, Harrison Young, Connor Walsh
Mentor: Conor Walsh

Highlight and Abstract: The field of soft robotics has garnered attention as a medium for education due to its hands-on fabrication methods, accessible materials, and applications in a wide range of STEM fields. With the potential for a more diverse set of disciplines and applications, a soft aquatic robot was developed for educational purposes as a project inspired by marine biologists' use of underwater robots that mimic fish to explore underwater habitats. The soft aquatic robot is simple, easy to build, and functional, making it an ideal candidate for use as a manufacturable educational tool for K-12 students. The aquatic robot is composed of a hybrid rigid body and a soft tail that uses the Fin Ray® Effect, a model of fish fin deformation during swimming, and is oscillated by two electromagnets located in the watertight body. The robot uses an open-loop control system to simplify the system's electronics. The use of an off-the-shelf microcontroller and motor driver to alternate the polarity of the electromagnets allows students to learn widely used tools to control robotic platforms, including Arduino microcontrollers. As well, the use of a Hall Effect sensor as a switching device opens a new world of science and creativity to students. The electronics are also designed to be simplistic and compact, allowing ease of setup and minimization of the robot's size. The replicability was measured as a percentage of successful replications of the electronics through user testing from the targeted demographic, K-12 school students, in an outreach program. There is also the element of programmability, which further allows students to begin to develop programming and computational skills. The soft aquatic robot was tested with a group of students to validate an increase in familiarity and interest in engineering fields. The development of this low-cost and accessible kit will allow young students to be exposed to new materials, robotics, and the field of engineering, which are less accessible due to high costs and lack of equipment.

Jaylynn Kim, Baran Mensah, Harrison Young, Lucas Gerez, Holly Golecki, Conor Walsh
Mentor: Conor Walsh

Highlight and Abstract: The field of soft robotics has garnered attention as a medium for education due to its hands-on fabrication methods, accessible materials, and applications in a wide range of STEM fields. With the potential for a more diverse set of disciplines and applications, a soft aquatic robot was developed for educational purposes as a project inspired by marine biologists' use of underwater robots that mimic fish to explore underwater habitats. The soft aquatic robot is simple, easy to build, and functional, making it an ideal candidate for use as a manufacturable educational tool for K-12 students. The aquatic robot is composed of a hybrid rigid body and a soft tail that uses the Fin Ray® Effect, a model of fish fin deformation during swimming, and is oscillated by two electromagnets located in the watertight body. The soft tail was designed to be compatible with a two part open-faced mold, facilitating the replication and manufacturing of the soft structure. The two part mold utilizes the deformation of plastic to create a break-apart structure contributing to the ease of release. The open-faced molding method eliminates the need for special equipment allowing for the fabrication of the tail in classroom settings. The body was modeled to have a generic shape with modular head and arm pieces that allow for customization of the aquatic robot. These pieces can further be developed to add functionality in terms of buoyancy and control. The replicability was measured as a percentage of successful replications of the tail through user testing from the targeted demographic, K-12 school students, in an outreach program. The soft aquatic robot was tested with a group of students to validate an increase in familiarity and interest in engineering fields. The development of this low cost and accessible kit will allow young students to be exposed to new materials, robotics, and the field of engineering, which are less accessible due to high cost and equipment.

Allie Gipson, Suji Choi, Kit Parker
Mentor: Kit Parker

Highlight and Abstract: Hydrogels are attractive materials for tissue engineering, but efforts to date have shown limited ability to produce the microstructural features necessary to promote cellular self-assembly into hierarchical 3D organ models. Using hydrogel ink containing prefabricated gelatin fibers is the state of the art for printing 3D organ level scaffolds. By using Dynamic Mechanical Analysis, we can characterize the stiffness of the scaffolds, with different scaffold fabricating conditions. We know that the mechanical properties of the substrate regulates various cell processes, like cell viability. Using live/dead cell analysis we find that a softer matrix will have better cell viability. As we build the scaffolds to better recapitulate the mechanical environment of the heart tissue the heart tissue/organ model built from FIG scaffolds will have physiological relevant heart performance.

Catherine Pfaltzgraff, Huibin Chang, Dr. Kevin Kit Parker
Mentor: Kit Parker

Highlight and Abstract: Cosmetic facial masks are projected to have an accelerated growth of popularity over the next decade. Typically, masks are made from non-woven materials which are not sustainable, however there is growing consumer need for sustainable personal care products. The main production methods of face mask come with the limitation of a low throughput production causing high cost, as well as low surface-to-volume ratio leading to poor skin absorption. Facial masks also lack the ability to be personalized based on different regions on the face. In this research we present the use of pullulan-based fiber, enhanced with glycerol, infused with the active ingredients such as lemongrass, thyme oil, and hyaluronic acid, in the production of a sustainable personalized face masks manufactured via Focused Rotary Jet Spinning (FRJS). The effects of glycerol and the active ingredients was examined in respects to the fiber composition, surface morphology, and mechanical properties. The active ingredients were added to a pullulan solution, enhanced with glycerol. The fibers were then spun using the FRJS. The fibers' composition was then analyzed using Fourier-transformed infrared spectroscopy (FTIR), revealing incorporations of the actives into the pullulan fibers. The morphology was examined by scanning electron microcopy (SEM), showing minimal effects of the active ingredients on the fiber's formation. The mechanical properties of the pullulan fibers were also evaluated by tensile testing. The results revealed higher tensile strength of the composite fibers as compared to pure pullulan fibers. The fibers containing the active ingredients were then spun on localized regions of the face mold to create a two layered mask, which consist of an active ingredient layer and a structural layer.

Christopher R. Warren, Michael M. Peters, K. Kit Parker
Mentor: Kit Parker

Highlight and Abstract: Focused Rotary Jet Spinning (FJRS) is used to engineer nano-fiber cardiovascular tissue scaffolds with varying degrees of anisotropy. This work shows that myocardial conduction velocity can be tuned by controlling scaffold organization. In vitro fiber-chips with fibers of isotropic, moderately anisotropic, and anisotropic configurations were produced and then “seeded” with neonatal rat ventricular myocytes (NRVMs). After five days in culture, the conduction velocities of these fiber-chips were calculated using a calcium-trace optical mapping system; the results indicate that the conduction velocity of engineered myocardial conduction is directly related to the degree of tissue anisotropy. These results characterize novel capabilities that set the stage for the engineering of cardiac tissues with high degrees of complexity.

Robert Lamarche, David Bell

Michael Bregar, Paul Chevalier, Federico Capasso
Mentor: Kit Parker

Highlight and Abstract: Generating terahertz frequency radiation has been a challenging problem in physics for several decades. In recent work, the Capasso Group at Harvard addressed the problem of frequency tunability by generating terahertz radiation using a quantum cascade laser (QCL)-pumped molecular laser (QPML) concept. A shortcoming of the current QPML cavity design is the output coupler. The goal of the project was to design a broadband output coupler for QPML terahertz lasers that had high efficiency and low losses in a wide frequency range by running simulations for different output coupler designs, materials, sizes, and thicknesses.

Simulations were performed with various output coupler designs such as periodic arrays (periodicity g) of metallic strips, squares, and circles. The circular design (radius r) was promising since it had a large region of frequencies with high reflection (≈90%) and low transmission (≈10%), which allows for amplification of the light inside the laser cavity by stimulated emission. For substrate materials with different optical indices (n), it was found that the reflection and transmission functions of the normalized frequency (defined as n g/Λ with periodicity g and wavelength Λ) over the region 0.017 to 0.95, were similar. The absolute operating frequency region of a given array thus depends on the substrate index of refraction. Different periodicities were simulated to find the frequency regions that maximized efficiency and reflection while minimizing losses. Although a wide range of g values were simulated, the reflection and efficiency were consistently high and the losses consistently low near n g/λ =0.8 and r/g =0.35, meaning that the coupler should be designed with a r/g ratio of approximately 0.35 for normalized terahertz wavelengths of Λ=n g/0.8. The new designed output coupler exhibited a higher efficiency (maximum around 60%) than the previous pinhole coupler, making it an attractive alternative coupler for terahertz lasing applications.

Anna Li, Danial Haei Najafabadi, Yue Luo, William Wilson
Mentor: William Wilson

Highlight and Abstract: Two-dimensional (2D) materials are extremely thin (1000x thinner than human hair) materials whose optical and electronic properties change drastically compared to bulk material. 2D materials have many applications as semiconductors and superconductors, as well as in newer fields like quantum computing. By selective stacking and manipulation of lattice orientation angle, we can engineer materials with new and exotic properties.

Anna Li, Danial Haei Najafabadi, Yue Luo, William Wilson
Mentor: Dr. Ling Xie

Highlight and Abstract: Silicon micropillar arrays, which have applications in microfluidics, optoelectronics, and biosensing, are usually fabricated using the Bosch process of deep reactive ion etching. We investigated how photolithography and etching conditions affect the final shapes of the pillars in order to optimize them. Here, we used a maskless aligner to directly write the micropillar arrays onto positive photoresist, and after development of the photoresist, we etched the substrate with varying conditions. We then characterized the arrays using scanning electron microscopy to measure critical dimensions and the etch rates of both the silicon and photoresist. Through this process, we established that there is an optimal exposure per micron of photoresist independent of array dimensions. In the etching process, we used low platen power that ramps down with time to avoid undercut, and in this way, we were able to fabricate pillars with diameters as small as one micron. We also found that with some of the etching conditions we studied, the etching process has an extremely high selectivity of well over 120:1 to Microposit S1800 series photoresist and that the etch rate of silicon accelerates with depth; the two effects are inexplicable yet well-observed.

Christina Perry, Kathryn Hollar
Mentors: Kathryn Hollar, Pia Sörensen

Highlight and Abstract: Nixtamalization is a traditional technique used to prepare maize, and it includes treating the maize with a basic solution consisting of ash, then steeping the maize in the basic solution known as nejayote, and then rinsing the steeped maize which is referred to nixtamal. The word nixtamalization comes from the Nahuatl words nexti, meaning ashes, and tamalli, meaning tamale. Several studies have explored the effect of nixtamalization on the nutritional content and texture of maize. This study aims to report the changes in corn kernels during each stage of the nixtamalization process, along with observing changes in niacin or vitamin B3 and other amino acids. Various types of corn were studied, including White Olotillo, Yellow Cónico, Pink Xocoyul, Purple Cónico, Blue Cónico, and Chico corn. Two nixtamalization methods were used to prepare the various corn types: the cook first method and the soak first method. Samples of corn were obtained and stored for analysis throughout each stage of performing the cook first and soak first methods. The size, weight, and color were the primary properties studied from the collected samples. Furthermore, Liquid chromatography-mass spectrometry (LC-MS) measured the nutrient content of the corn. The corn kernels' size and weight increased significantly after undergoing nixtamalization. The data collected from the LC-MS does not demonstrate an increase in niacin within the final product.