2021 Graduate Student Symposium

Help us judge student presentations!

Please fill out this form to help us score the presenters in Student Presentation Session I/II.

Multivariate Analysis of Peptide-Driven Nucleation and Growth of Au Nanoparticles

Kacper Lachowski

Sequence defined molecules such as peptides and peptoids gives researchers significant control over the formation of inorganic nano-structured materials. Bio-inspired synthesis and assembly approaches have been applied to noble metals, perovskites, quantum dots, titania, and other systems. Frequently, the use of sequenced defined molecules in inorganic nanoparticle synthesis is motivated by enabling the formation of complex structures at mild reaction conditions. However, predicting the effect of sequence defined molecules on inorganic material synthesis is an ongoing issue. Small changes to the identity or the order of a molecule’s sequence can have a strong impact on how it interacts with the inorganic precursors or particles in a reaction (20n unique peptide sequences for peptide of length n). Moreover, molecular design rules also need to account for factors other than molecular sequence when comparing the effects of molecules on synthesis outcomes. To address this problem, we focus on the use of Au binding peptides in the synthesis of Au nanoparticles. We created 6 variants of a Au binding peptide (Z2) and studied how each peptide impacted the synthesis of Au nanoparticles in 64 reaction conditions per peptide by varying the concentration of precursor, reducing agent, and peptide. Each sample was characterized by UV-Vis spectroscopy which serves as a proxy for the structure of plasmonic nanoparticles. By applying functional data analysis to these measurements, we were able to make comparisons between the 6 peptide variants in a large experimental design space.

Developing antibody-dependent CRISPRa in cell-free systems

Ryan Cardiff

With the emergence of new, highly contagious SARS-Cov-2 variants, the need for rapid and accurate antibody testing remains critical. In this work, we highlight progress toward developing a CRISPR based system in which fluorescent readout is dependent on the presence of one or more specific antibodies in an E. coli based cell-free system. The inherent multiplexing capabilities of CRISPR-activation (CRISPRa) enables the development of new diagnostics that are able to process information from multiple inputs and produce signals only upon detection of multiple antibodies, resulting in lower-false positive rates. In this system, the dCas9-gRNA complex and transcriptional activation domain are both recruited to the antibody of interest through the fusion of antigen peptide tags or IgG-binding proteins to the CRISPRa machinery. Through this antibody-mediated linkage, the dCas9 complex is able to recruit the activator to a reporter gene and initiate expression. To facilitate antibody-dependent CRISPRa, we develop a synthetic amplifier with improved performance and characterize more complicated CRISPRa architectures to deepen our understanding of the rules governing activatability. When successful, this approach will enable rapid and quantitative point-of-care antibody testing.

Machine Learning Assisted Enhancement of Perovskite Solar Cell Stability and Performance

Preetham Paul Sunkari

Hybrid perovskites, or simply perovskites, have gathered widespread attention among the solar cell research community due to their exceptional optoelectronic properties and a rapid increase in their power conversion efficiencies from 14% to 25.2% in just a span of seven years. Despite being cheap, easy to manufacture, and highly efficient, perovskite solar cells have not entered the commercial market due to their unpredictable stabilities when exposed to oxygen, moisture, heat, and light. Moreover, until now, there has been no consensus in the research community on appropriate stability tests that can accurately determine if a perovskite solar cell can last long enough in operational deployment. In the Hillhouse group, we developed an accelerated protocol for material-level assessment of hybrid perovskites under different operational stresses. We are the first to demonstrate experimental machine learning models that predict the performance lifetimes of these materials using physics-informed features derived from early-time optoelectronic measurements like photoconductivity, wide-field photoluminescence, and thin-film transmittance. Our feature development efforts have also paved the way for developing an accurate kinetic model that describes the rate of degradation in the most popular perovskite material, methylammonium lead iodide (MAPbI3). Here, I present the various experimental and modeling techniques we employed for accelerated material (and device) stability evaluation and forecasting.

Urea Oxidation Performance and Stability In Saline Solutions of Commercially Available Alloys: BC1, C276, and C2000

Anthony Pyka

Production of economic, green energy has been in high demand as the efficiency of manufacturing processes have skyrocketed in the last few decades. Fuel cells, specifically hydrogen fuel cells, are attractive because of their high-power output with the main byproduct of water. Hydrogen generation through water electrolysis is a safe and go-to solution, but the coupled Oxygen-Evolution Reaction (OER) is sluggish. Here we replace the OER with the Urea Oxidation Reaction (UOR) as UOR requires a lower operating potential and has faster kinetics than OER.

Understanding lithium sulfur battery chemistry using fundamental thermodynamics, kinetics, and transport models

Linnette Teo

Lithium sulfur is a promising next-generation battery chemistry as electric transportation increases in scale. The progress of lithium sulfur development can be accelerated by sophisticated physics-based models. This poster will cover how we can build up our knowledge of a complex electrochemical system by starting with a thermodynamic model and incrementally adding physics.

Building CRISPRa/i Circuits through the Regulated Expression of Guide RNAs in Cell-Free Systems

Diego Alba

CRISPR-Cas transcriptional networks hold great promise as platforms for engineering metabolic pathways and information processing circuits. Historically, prokaryotic CRISPR control systems have been limited to CRISPRi. Creating approaches to integrate CRISPRa for transcriptional activation in these networks would greatly expand CRISPR circuit design space. In this work, we have developed design principles for engineering prokaryotic CRISPRa/i genetic circuits with network topologies specified by guide RNAs. We present multi-layer CRISPRa/i activation cascades and feed-forward loops operating through the regulated expression of guide RNAs in cell-free expression systems. The correspondence between predicted effects of CRISPRa/i circuit tuning actions and measured functions suggests that these circuits could be assembled into larger and more complex networks. We show that CRISPRa/i networks can be rationally tuned to program complex functions by designing type 1 incoherent feedforward loops (I1-FFL), which serve as tunable pulse-generators. This work establishes a framework for building scalable CRISPRa/i networks as regulatory programs in cell-free expression systems with applications in bioproduction and biocomputation.

Protein-based Hybrid and Hierarchical Nanomaterials

Nada Naser

Research in the Baneyx Lab lies at the interface of microbiology, nanotechnology, molecular engineering and materials science. We seek to understand the rules that underpin the interactions of solid-binding peptides with inorganic and synthetic interfaces. We genetically install peptides of different functionalities and binding affinities within the framework of protein scaffolds to fabricate stimuli-responsive hybrid systems and self-organizing living materials. Applications for hybrid materials range from biomedicine and bionanoelectronics to biomineralization and catalysis. We are a lead lab in CSSAS, the Center for the Science of Synthesis Across Scales, an Energy Frontier Research Center (EFRC) directed by Prof. Baneyx that brings together researchers from three universities and two national labs.

Machine learning partition functions for the acceleration of reaction kinetics computation

Evan Komp

Evaluating chemical reaction rate constants, k(T), is necessary to model reactive systems such as combustion, waste-water treatment, energy storage, among others. Many ab initio methods have been developed to compute k(T), ranging from transition state theory to dynamic approaches such as the flux-flux correlation function. Most require knowledge of either reactant and/or transition state partition functions (Q(T)). In the canonical ensemble these describe the reactive molecule’s occupation of energy levels as a function of temperature. Electronic structure theory methods are used to compute molecular energies and frequencies to be used for partition function calculations. For high accuracy electronic structure theories, the computational cost to obtain energies and frequencies may scale exponentially with the system size. To overcome this cost we used supervised machine learning, which has seen recent success in material science, to predict molecular partition functions as well as partition functions of unknown transition states.

Route-Specific Energy Requirements for Battery-Electric Buses, an Open-Software, Physics-Based Approach

Erica Eggleton

As transit agencies, such as King County Metro, transition to electrified vehicles, there are many new parameters that must be considered in route planning, fleet maintenance, and charging logistics. One major knowledge gap is the understanding of the lifetime of the energy storage system, such as lithium-ion battery packs, on-board. This project aims to create a software tool that these agencies can use to make logistic decisions that are predictive rather than reactive, using data that in most cases is already available. This software uses geographical information services (GIS) data, ridership data, and bus specifications to estimate the load and energy requirements for specific routes. Currently, this software is focusing on the systems level by providing insights into which routes are harder on the batteries and where to place charging stations to best optimize operation time. This talk will also briefly introduce how additional software modules can be used at the battery pack and single cell level to better understand degradation and lifetime of the energy storage system.

Effects of zwitterionic polymer brush density and chain length on resisting protein adsorption for biomaterials

Julia King

For implantable medical devices, such as insulin catheters, prolonging device lifetime is critical. For instance insulin catheters fail after 2-3 days due to occlusion by cell and collagen accumulation. To reduce the extent of occlusion, we have developed a nonfouling coating to resist protein adsorption—the suspected root cause of the failure. Our coating is composed of a surface-grafted polymer brush of sulfobetaine methacrylate (SBMA)—a super-hydrophilic and nonfouling material—and a radio frequency plasma-deposited polymerization initiator layer which allows for a geometry-independent, stable coating for polymeric biomaterials. The SBMA is polymerized using activators regenerated by electron transfer (ARGET) atom transfer radical polymerization (ATRP). In this work, we have investigated the effects of brush density and chain length in minimizing protein adsorption. We suspect that these factors should be balanced to produce a nonfouling coating that is the most sterically-favorable for maximum water binding. X-ray photoelectron spectroscopy (XPS) quantified the brush density relative to polymerization initiator surface concentration. An in vitro radiolabeled protein adsorption assay determined the effectiveness of the SBMA coating. Additionally, sum frequency generation (SFG) vibrational spectrometry was used to probe the water structure at the coating interface to comment on nonfouling character. In this work we have found the thickness of the SBMA film to be tunable with salt concentration added during the polymerization, lowering the critical solution temperature and enabling longer SBMA chain fabrication. We have shown for increased brush density, protein adsorption was lowered. SFG signals showed a larger tightly-bound water-to-loosely-bound water ratio for higher brush density surfaces than for lower brush density surfaces which is consistent with our protein adsorption results. In future work, we will push the fabrication limits of polymer brush density with the solubility limits of increasing chain length to achieve minimal protein attachment and ultimate nonfouling for implantable medical devices.