SEAS PhD Research Symposium

Applied Physics and Applied Mathematics

Optimization of flood mitigation in New York: An Adjoint Based Approach

Hurricanes threaten New York and other coastal cities primarily through coastal flooding known as storm surge. We can model storm surge for a given storm with shallow water equations. However, accurate predictions are highly computationally expensive. Flood mitigation strategies, such as sea wall placement, are often done in an improvised way since simulating damage for all possible wall locations and possible storms is intractable. An effective search for the optimal mitigation strategy requires a deeper understanding of how the dynamics of the water flow changes with properties of the storm and with parameters we want to change such as topography and ground roughness, In my research, I employ the adjoint-state method that allows us to find the sensitivity of a damage functional to these different parameters. This allows us to utilize descent methods which can reach a solution with fewer repeated simulations than a derivative-free search of the parameter space. In this way, we treat flood mitigation in New York City as a PDE-constrained optimization problem, but also seek to develop a robust theory that can be applied to any coastal city.

Biomedical Engineering

Common Neural Mechanisms Underlie Contralateral and Ipsilateral Negative BOLD Responses in Human Visual Cortex

The task-evoked positive BOLD response (PBR) due to a unilateral visual stimulus is often accompanied by sustained contralateral as well as ipsilateral (relative to the stimulus) negative BOLD responses (NBRs) in human visual cortex. Little is known about the origins and characteristics of these two types of NBRs. Hence, in this study, we used a unilateral visual task to stimulate human visual cortex during fMRI acquisition to investigate the properties of the two task-evoked NBRs. Using 42 healthy right-handed participants, we first demonstrated that the magnitudes of both NBRs linearly increase with the stimulus duration evidence for rough linearity. Then, we extracted the shape of the hemodynamic response functions (HRFs) of the two NBRs and demonstrated that the two are closely similar in both overall shape and timings. However, the dynamic of the HRF for contralateral NBR (cNBR) and ipsilateral NBR (iNBR) were significantly different from the PBR. While both NBRs showed a delayed on-set time, they reached their peaks faster, and also fell back to baseline quicker than PBR. Additionally, we showed that the subject-wise amplitude of the cNBR is tightly associated with that of iNBR which is much higher than their relationship with PBR. Furthermore, we showed that unlike PBR, neither of the NBRs were correlated with task performance. Finally, we demonstrated that attention modulates both NBRs differently from its effect on PBR. Our findings suggest that the visually evoked cNBR and iNBR are most likely generated throughout a common underlying neural and/or vascular mechanisms.

Task-Evoked Negative BOLD Response in the Default Mode Network Does Not Alter Its Functional Connectivity

While functional connectivity networks are often extracted from resting-state fMRI scans, they have been shown to be active during task performance as well. However, the effect of an in-scanner task on functional connectivity networks is not completely understood. While there is evidence that task-evoked positive BOLD response can alter functional connectivity networks, particularly in the primary sensorimotor cortices, the effect of task-evoked negative BOLD response on the functional connectivity of the Default mode network (DMN) is somewhat ambiguous. In this study, we aim to investigate whether task performance, which is associated with negative BOLD response in the DMN regions, alters the time-course of functional connectivity in the same regions obtained by independent component analysis (ICA). ICA has been used to effectively extract functional connectivity networks during task performance and resting-state. We first demonstrate that performing a simple visual-motor task alters the temporal time-course of the network extracted from the primary visual cortex. Then we show that despite detecting a robust task-evoked negative BOLD response in the DMN regions, a simple visual-motor task does not alter the functional connectivity of the DMN regions. Our findings suggest that different mechanisms may underlie the relationship between task-related activation/deactivation networks and the overlapping functional connectivity networks in the human large-scale brain networks.

Proteus Syndrome Modeling via CRISPR Gene Editing and Over-Expression

The lack of samples for rare diseases such as proteus syndrome has impeded the understanding of pathogenesis and providing effective treatments for those patients. In this work, we utilized the technology of CRISPR editing in human induced pluripotent stem cells (iPSCs) and endothelial cells to generate the AKT E17K mutation, and lentiviral systems to over-express the AKT E17K mutation that induces the proteus syndrome to model the overgrowth of the patient tissues and malfunctions in the micro vessels and brain. Combining the methods of gene editing and organ-on-a-chip, we are aim to understand the malfunction of the angiogenesis for the AKT E17K mutated micro-vessel and reveal the pathogenesis of the proteus syndrome and provide potential drugs for further clinical trials.

Oral Delivery of Genome Editing Components for Ionizing Radiation Protection

High-dose exposure of ionizing radiation to body can lead to acute radiation syndrome (ARS), causing significant damage to the gastrointestinal (GI) system and hematopoietic (HP) cells. To avoid multiple injections and invasive surgeries, gene editing may protect and regenerate intestinal crypt cells and HP cells, thus enhancing the resilience and survival of intestinal and hematopoietic cells upon ionizing irradiation. In this research, we aim to develop an effective oral delivery system for the nonviral gene-editing components. We used a two-step process to synthesize chitosan-grafted branched polyethylenimine (CS-g-bPEI). These cationic polymers could complex with CRISPR/Cas9 components to form nanoparticles (NPs) for intracellular delivery. The size (~80 nm) and zeta potential (~55 mV) enabled CS-g-bPEI/pDNA NPs to cross gut epithelium and enter intestinal and hepatic cells through transcytosis and endocytosis. Cas9 and sgRNA were transfected in human colorectal and hepatocellular carcinoma cell lines, outperforming the transfection reagent Lipofectamine 3000 without significant cytotoxicity. The G-CSF gene activation efficiencies are currently being investigated in HCT 116 and Hep G2 cell lines. We are also developing a flash nano-complexation (FNC) technology for NP coating and scale-up production. This work offers a versatile oral nonviral genome editing platform that can meet various existing or emerging medical needs in the future. 


Deep Learning for Sustaining Human Eyesight: Automated Glaucoma Detection from OCT Images

As one of the leading causes of irreversible blindness worldwide, glaucoma treatment costs the U.S. health care system alone an estimated 2.5 billion dollars annually. Although methods to detect and to slow the progression of the disease exist, one of the biggest challenges is facilitating timely and accurate screening via regular ophthalmology exams. Deep learning has the potential to play a major role in automating early screening of glaucoma by indicating to patients their extent of glaucoma risk, ensuring earlier visits to specialists to receive eyesight-preserving treatment. We describe and assess convolutional neural network (CNN) models for their ability to detect glaucoma from optical coherence tomography (OCT) retinal nerve fiber layer (RNFL) probability maps as well as a combination of other OCT-derived images. CNNs pretrained on natural images performed comparably to CNNs trained solely on RNFL probability maps, and models showed high accuracy on the order of 95%. Class activation maps highlighting regions of interest chosen by CNNs to make classification decisions suggest methods to improve accuracy and reduce false positives and false negatives, such as incorporating additional multi-modal information contained within OCT reports into model design and training. Such deep learning systems have the capacity to work in tandem with human experts to maintain overall eye health and expedite detection of eye disease, enabling a more sustainable health care system and a more independent, sighted population across all ages.

Biological Nanowires: Silver(I)-Mediated Base Pairing in DNA Nanowires and Nanoscale Networks

Recent advances in DNA chemistry have isolated and characterized an orthogonal DNA base pair using standard nucleobases: by bridging the gap between mismatched cytosine nucleotides, silver(I) ions can be selectively incorporated into the DNA helix with atomic resolution. The goal of this work is to explore how this approach to “metallize” DNA can be combined with structural DNA nanotechnology as a step toward creating electronically-functional DNA networks. We investigate the assembly of linear Ag+-functionalized DNA species using biochemical and structural analyses to gain an understanding of the kinetics, yield, morphology, and behavior of this orthogonal DNA base pair. Then we carry out scanning tunneling microscope (STM) break junction experiments on short polycytosine, polycationic DNA duplexes and find increased molecular conductance of at least an order of magnitude relative to the most conductive DNA analog.

With an understanding of linear species from both biochemical and nanoelectronic perspectives, we investigate the assembly of nonlinear Ag+-functionalized DNA species. Using rational design principles gathered from the analysis of linear species, a de novo mathematical framework for understanding generalized DNA networks is developed. This provides the basis for a computational model built in Matlab that is able to design DNA networks and nanostructures using arbitrary base parity. With this foundation, three general classes of DNA tiles are designed with embedded nanowire elements: single crossover Holliday junction (HJ) tiles, T-junction (TJ) units, and double crossover (DX) tile pairs and structures. A library of orthogonal chemistry DNA nanotechnology is described with applications to nanomaterials and circuit architectures.

In vivo proton spectroscopy evidence of differential glutathione metabolism in relapsing-remitting and progressive multiple sclerosis frontal cortex

We applied 7 Tesla proton magnetic resonance spectroscopy to compare frontal cortex concentrations of glutathione, γ-amino butyric acid (GABA), glutamate, and five additional compounds in individuals with relapsing-remitting (13 female, 6 male; mean 50 ± standard deviation 10 years old), progressive (12 female, 9 male; 55 ± 8 years), and no (9 female, 7 male; 52 ± 10 years) multiple sclerosis. Progressive multiple sclerosis patients exhibited reductions in glutamate +glutamine concomitant with decreases in glutamate but not glutamine, while relapsing-remitting patients demonstrated increases in glutathione precursor glycine. Gradient-boosted binary logistic regression identified glutamate and glycine, respectively, as the most important metabolites for modeling either progressive or relapsing-remitting multiple sclerosis. Relapsing-remitting patients exhibited an abnormal positive correlation between glutathione and glutamate + glutamine, while progressive patients expressed an abnormal negative correlation between glutathione and myoinositol. The progressive group exhibited a trend to tissue-specific abnormalities in glutathione, which demonstrated a positive correlation with tissue volume in control that significantly differed from null relationships in both disease groups. Both disease groups exhibited age-dependent GABA concentration abnormalities with excitatory-inhibitory neurotransmitter imbalance. Our results constitute the first in vivo evidence of differential contribution by glutathione metabolism to relapsing- remitting and progressive multiple sclerosis pathology.

Chemical Engineering

3D Patterning of Nanoparticles by Molecular Stamping

Molecular patterning of nanoparticles can allow prescribing inter-particle binding sites, and that can permit creating tailored building blocks for assembly of targeted nanoparticle clusters and large-scale periodic nano-organizations. However, establishing a platform for rational pattering of nanoparticles has been challenging. We have developed a so-called Molecular Stamping approach for 3D nanoparticle patterning by templating target particles with designed molecular patches using DNA frame. In this method, molecular stamps, located at the prescribed locations of frame, can be engaged for transferring distinctive molecules onto different sites of nanoparticle surface. Our approach allows for creating molecular patches with a single molecule control and with same or different binding affinities. We applied the developed method to generate planar and 3D arrangements of molecular patches on nanoparticles, and investigated factors affecting pattern formation and its resolution. The developed approach was used to create nanocluster architectures containing nanoparticles of different sizes and from different materials. Electron microcopy and tomography were utilized to reveal spatial arrangements of molecular patches and the factors contributing to the patterning process.


Civil Engineering and Engineering Mechanics

Design and Demonstration of a Self-powered Wireless Sensor Network Using the Thermoelectric Cells in Window Frames

A self-powered wireless sensing network (SPWSN) unit is designed and prototyped for the thermal energy harvesting on the window system. The proposed system contributes to the smart building design and optimizes the energy consumption of the building sector, and therefore can significantly improve the sustainability of engineering. The system is designed with the energy harvesting unit and the wireless sensing network unit. In the energy harvesting unit, the thermoelectric generators are used to extract energy from the thermal gradient in the window system. The LTC3108 and LTC4071 are used as the voltage boosting and battery management integrated circuits for energy storage. The total power harvested reaches 1.5 mW from 4 TEGs under 5.5 controlled temperature difference, and the energy efficiency of the system reaches 33.4%. The wireless sensor network unit is designed based on the ESP 32 microcontroller and integrated with the temperature and moisture sensor. The data extracted is wirelessly transmitted to the cloud through Wi-Fi and serves as valuable information for the design and optimization of the building information system. The field tests inside the window frame are conducted, and the energy balance is achieved between energy harvesting and consumption. The energy equilibrium algorithm is proposed based on the test result, and the battery energy level can be projected given the historical weather information, which serves as an important reference for the design and engineering.

Optimal Coastal Protective Strategy Against Storm Surges and Sea Level Rise

Interdependent critical infrastructures including transportation, the power-grid, and emergency services in coastal regions are threatened by storm-induced flooding. Massive hurricane events such as Hurricane Sandy and Katrina have demonstrated the need for plans to protect our infrastructure. On the other hand, the events only reflect a possible future threat and do not fully address the unknown probability and impacts of all possible future threats. This uncertainty is only amplified by climatic effects such as sea-level rise.

The project goal addresses the threat of storm-induced flooding to interdependent critical structures by developing a methodology that can search for optimal adaptation strategies for a sustainable future. The proposed methodology, including flooding simulation on GeoClaw / Geographical Information System (GIS) and damage assessment on GIS, would optimize strategies to maximize their protective abilities over time and space constrained by budgetary considerations and stakeholder’s observations for the interdependency of infrastructures. GIS also provides damage cost estimates that include building physical damage cost, inventory loss, and income loss due to hurricanes. The optimizer will also use the GIS-based model in the first iterations of the optimization in order to save time and only use the more expensive GeoClaw based simulation when getting closer to an optimal strategy.


Design of a Geothermal Well Filled with Phase Change Materials for Daily and Seasonal Heat Storage and Supply

Geothermal heat pumps have almost no negative effects on the environment, and even have positive effects as they reduce the usage of other environmentally unfriendly energy sources.

This project aims at designing a geothermal well for daily and seasonal heat storage and supply for energy efficient buildings as well as other geothermal applications. Coupling with heating, ventilation, and air conditioning (HVAC) systems and/or building integrated photovoltaic thermal (BIPVT) systems, it can not only significantly boost the efficiency HVAC and BIPVT systems, but also be used for inter-seasonal heat exchange of heating (winter) and cooling (summer) for energy efficient buildings.

Inclusion-based Boundary Element Method for Design of PCM Embedded Building Materials

The embedment of microencapsulated phase change materials (PCMs) is a promising means to improve the thermal inertia of building materials. However, PCMs are often soft inclusions with wide particle size distribution, which degrades mechanical properties and complicates the heat transfer behavior of the composite. This poster establishes a new approach to simulate inhomogeneities introduced by PCM addition by using Green’s function to determine the thermal fields of cementitious composites containing PCMs. This inclusion-based boundary element method (iBEM) is mesh-free and able to capture heat conduction in multi-component particulate composites. The reliability of this model is validated by realistic experiments on thermal response of wall panels PCMs. The iBEM model is further applied to develop a design rule for performance equivalence through parametric study, such that the energy efficiency of a building envelop can be maintained equivalent to that with R values required by building codes.

Design and Demonstration of a Hybrid Geothermal-Solar Heating and Cooling of Pavement Overlays

Weather can negatively affect the structural integrity of pavement and reduce its efficacy and safety. Low temperatures in winter could cause thermal cracking due to thermal tensile stress in different layers. Pavement can also undergo thermal-fatigue cracking which occurs when thermal fatigue distress due to daily temperature cycling exceeds pavement fatigue resistance. This technology describes a hydronic pavement system that addresses issues occurring for both hot and cold climates using sustainable energy sources. Pipes filled with fluid are embedded within the pavement that can either dissipate heat into the ground (for summer) or extract heat from geothermal pumps to regulate temperature (for winter). The pavement is fitted with sensors that can automate the temperature control of the system, where the sensors and pumps are powered by solar panels. The asphalt or concrete will be modified with carbon nanotubes or graphene nanoplatelets in order to have a higher thermal conductivity. As a result, this system can be used in different types of climates to improve safety and comfort, and reduce maintenance need.

Thermal Strain and Cracking Analysis of Layered Composites towards the Design of Solar Blinds

Solar windows are fabricated by attaching solar cells to the window blinds, generating electricity when facing the sunlight. Window slats can be made of glass-cell-glass (GCG) layers, or glass-cell-Tedlar (GCT) layers bonded by ethylene-vinyl acetate (EVA). Due to the different thermal expansion coefficients, layered composites may exhibit significant thermal stress and deformation when subjected to a temperature change. The two types of slats behave differently during temperature change: the GCT shows curling, whereas the GCG keeps straight due to the symmetry in geometry and material property. Obviously, both shearing and bending mechanisms exist in general layered materials, which can be explained by the shear lag model and the classical beam theory, respectively. Nevertheless, a unified theory is needed to account for both mechanisms. This paper proposes a novel formulation considering both mechanisms through the principle of stationary potential energy. A perfectly bonded interface is assumed, and the boundary value problem is formulated and solved theoretically. Numerical verification is conducted by comparing with the ABAQUS simulations and good agreements are achieved. The formulation is extended to the fracture analysis with the virtual crack extension method (VCEM) and the fracture pattern can therefore be predicted. The parametric analysis is conducted to estimate the influence of parameters for better design and engineering.

Earth and Environmental Engineering

Advancing Conductivity-Permselectivity Tradeoff of Ion-Exchange Membranes with Sulfonated CNT Nanocomposites

Ion exchange membranes (IEMs) are widely applied in energy and water technologies, such as reverse electrodialysis for sustainable power generation and electrodialysis for desalination. An IEM with lower ionic resistance and better selectivity for counterions can reduce the energy demand and raise the efficiency of IEM-based processes. However, an empirically-observed tradeoff between conductivity and permselectivity always constrains the performance of IEMs. The incorporation of rationally functionalized 1-dimensional nanomaterials as filler into the polymer matrix offers opportunities to depart from this tradeoff. When highly dispersed, the 1-D nanomaterials with large aspect ratio can form a percolating network to facilitate ion transport. In this study, we developed nanocomposite cation exchange membranes (CEMs) by incorporating sulfonic acid-functionalized carbon nanotubes (sCNTs) in sulfonated poly(2,6-dimethyl-1,4-phenyleneoxide), sPPO, polymer matrix. The fabricated nanocomposite IEMs exhibit improved conductivity while maintaining permselectivity. Intrinsic resistivity, the reciprocal of conductivity, is lowered with greater blending of sCNTs filler into the sPPO matrix, decreasing by approximately 25% with 20 w/w% incorporation of sCNTs. The variation of permselectivity is within 3% across the different degrees of sCNT incorporation. Therefore, compared with pristine membranes, the conductivity-permselectivity tradeoff line of the fabricated nanocomposite membranes are advantageously advanced, improving overall performance. Enhancement in conductivity is more pronounced for membranes with lower swelling degree. We posit that the interconnected network of sCNT reduces the tortuosity of the ion diffusion path, thereby increasing effective ionic conductivity. This study demonstrates the potential of rational utilization of nanomaterials to advance the conductivity-permselectivity tradeoff governing conventional IEMs.

Electrical Engineering

Enhancement mode ion-based transistor – comprehensive interface for in vivo electrophysiology

Bioelectronic devices must have high speed and high performance to interact with the often rapid, low amplitude signals generated by neural tissue. These devices should also be biocompatible, soft, and exhibit long-term stability in physiologic environments. Enhancement mode transistors are critical components of such devices, but current architectures do not combine these properties. Here, we develop an enhancement mode ion-gated transistor (e-IGT) that functions based on a reversible redox reaction and hydrated ion reservoirs within the conducting polymer channel to enable long-term stable operation and shortened ion transit time. E-IGTs have high transconductance and fast transient responses, resulting in a gain bandwidth product that is several orders of magnitude above other ion-based transistors. We usesd these transistors to acquire a wide range of electrophysiological signals, including in vivo recording of neural action potentials. E-IGTs offer a safe, reliable, and high performance building block for chronically implanted bioelectronics, with spatiotemporal resolution to the scale of individual neurons.

Mechanical Engineering

Combining Physics and Data across Multiple Scales in Chemically Reacting Flows

All chemically reacting flows are inherently multi-scale systems, meaning that interactions among nuclei and electrons at the molecular level govern chemical reactivity at the macroscopic level. Understanding these interactions across all scales is imperative for developing accurate chemical kinetics models that can be used for predictive simulations. Embracing the multi scale nature of chemically reacting flows enables one to employ powerful strategies for uncertainty quantification and informatics as well as quantifying non-equilibrium kinetics in realistic mixtures. These strategies allow us to leverage improved energy calculations, due to the recent progress in theoretical and computational methods, and consider theoretical and experimental data across multiple scales on equal footing in the same mathematical framework. This is known as the Multi-scale Informatics (MSI) approach. The Multi-scale informatics approach works to identify optimized values and quantified uncertainties for a set of theoretical kinetic parameters that describe the overall kinetic model in a manner informed by available data from ab initio electronic structures, measured global observables, and rate constant determinations. It is our hope the optimized kinetic models resulting from the MSI approach can be used to accelerate design of things like new transportation engine concepts with better fuel economy or accelerate the development and understanding of new bio-fuels.

Passive Radiative Cooling By PDMS Selective Emitters

The scalability and implementation of selective emitters in passive radiative cooling applications are limited by the high fabrication costs due to the complexity of photonic structures. The usage of commercially available polymers in selective emitters holds potential in lowering the cost of radiative cooling solutions. In this work, we demonstrate that thin films of polydimethylsiloxane (PDMS) on aluminum substrates act as radiative coolers by selectively emitting in the wavelength range of 8 µm to 13 µm, where the Earth’s atmosphere is highly transparent. We also show that our device can achieve passive cooling up to 12 C below the ambient temperature under the night sky. This suggests that PDMS, especially due to its ease of deposition, may be a viable selective emitter in passive radiative cooling applications.