iSURE 2014 Projects

Project: Synthesis and Catalysis of Bimetallic Nanocatalysts Faculty name and email: Dr. Franklin (Feng) Tao, ftao@nd.edu ND College: College of Science Project description: Research of Franklin (Feng) Tao group is in the interdisciplinary field of heterogeneous catalysis and nanoscience for efficient energy conversion. We focus on nanocatalysis crucial for efficient energy harvest and conversion, chemical transformation, and environmental remediation. The goal of our research projects is to develop efficient nanocatalysis using different syntheses that build on in-situ operando studies of nanocatalysis using new analytical techniques including in-house (using Al Kα) ambient pressure X-ray photoelectron spectroscopy available in our group. Our research activity includes synthesis of nanocatalysts, measurement of catalytic performance and energy efficiency, and in-situ characterization under reaction conditions. We currently work on synthesis and catalysis in generation of hydrogen from methanol steaming reforming and partial oxidation, conversion of carbon dioxide to fuel molecules through thermal catalysis, selective production of transportation fuels from syngas, generation of hydrogen from water through photocatalysis, and catalysis toward high selectivity in environment remediation. In addition, we collaborate with a few groups by using multiple in-situ techniques including environmental TEM (ETEM) and extended X-ray absorption fine structure (EXASFS) toward a deep and fundamental understanding of nanocatalysts at molecular and atomic. Pre-requisite courses or skills: Experience in synthesis of nanomaterials with colloidal chemistry is a plus. Project: Development of catalysts for the activation of Carbon Dioxide Faculty Name and email: Dr. Ken Henderson, khenders@nd.edu ND College: College of Science Project description: Polycarbonates are an economically valuable class of polymer, yet the traditional synthesis of these materials utilizes phosgene and produces toxic gases making this process environmentally unfriendly. More recent advances have been able to utilize ‘waste’ carbon dioxide; however, these processes often use environmentally and biologically harmful metals (Co, Cr, Sn) and fail to achieve high turn-over numbers (TONs). This would be very attractive from an energy perspective as CO2 generated by fossil fuel utilization could become a useful starting material for a value added product. Thus, this would help offset

some of the costs for CO2 separation processes, in turn making these more economically viable. We have initiated a new project in our laboratory to investigate the use of s-block metals (Li, Na, K, Mg, Ca) as a replacement for the mentioned toxic and costly metals. The light s-block metals are biologically and environmentally begin; additionally, these metals are earth-abundant, making them cost effective alternatives. In addition, these metals offer a gradated palette of physical and electronic properties, allowing tremendous control over the fine-tuning of the catalysts. Pre-requisite courses or skills: None Project: Hybrid MPI and CUDA Parallel Programming for Simulating Blood Flow Faculty name and email: Dr. Zhiliang Xu, zxu2@nd.edu ND College: College of Science Project description: The project is to use a hybrid programming model, i.e., MPI + CUDA, to parallelize the existing simulation codes for simulating platelets and other blood cells interacting with blood flow. The current simulation codes use CUDA to achieve parallelism. However, they only run on a single GPU device. Thus the size of problems that we can simulate is limited by the available GPU memories. We will utilize MPI + CUDA to build multi-level parallelism so that the simulation codes could run on multiple CPUs + GPUs to achieve massive parallelism. Pre-requisite courses or skills: 1. Numerical analysis (specially numerical differential equations) 2. Scientific Computing (using C or C++) 3. Knowledge of CUDA would be good, however it is optional Project: Nonclassical Oxygenation Reactions Faculty name and email: Dr. Seth N. Brown, Seth.N.Brown.114@nd.edu ND College: College of Science Project description: Activation of molecular oxygen and other oxidants for the oxidation of organic substrates remains a challenge to chemists. We are pursuing a strategy that involves using transition metals to stabilize bonds to oxygen, while the organic ligands on the metals store oxidizing equivalents, as a "nonclassical" way to solve this problem. This summer project would involve (1) preparing metal complexes suitable for achieving these oxidations; (2) surveying the scope of the reactions; and (3) conducting mechanistic experiments to probe the degree

of metal-ligand communication in such reactions. Pre-requisite courses or skills: Classroom laboratory experience with synthetic organic or inorganic chemistry Project: Synthesize Hyperbranched Polymers with Uniform Structure – Mechanistic Study Faculty name and email: Dr. Haifeng Gao, hgao@nd.edu ND College: College of Science Project description: Recently, a new synthetic strategy is developed in our research group that successfully produces hyperbranched polymers with high molecular weight (Mn ~ 106) and uniform structure (Mw/Mn ~ 1.2) using a one-pot microemulsion polymerization technique. The initial success reveals a general strategy of using a discrete confined space to produce functional hyperbranched polymers with tunable molecular weights and hierarchical structures. The current project is to fully investigate the features of this robust technique and to expand its application to obtain functional hyperbranched polymers for potential application in site isolation and catalysis. Pre-requisite courses or skills: Polymer synthesis and characterization Project: Stochastic computing and nanomagnet logic (NML) Faculty name and email: Dr. Sharon Hu, shu@nd.edu ND College: College of Engineering Project description: By placing nano-scale magnets in carefully crafted patterns, logic computation can be performed. Such nanomagnet logic (NML) circuits provide a drastically different way of processing data from traditional CMOS. NML circuits have many desired properties, including lower power, non-volatility, and radiation hardness. Basic structures of NML circuits have been experimentally demonstrated. NML circuits, however, are fundamentally more error prone than charge-based devices. Stochastic computing employs bit streams that encode probability values to represent and process information. If designed properly, a small number of bit flips (regardless of their position) in a long bit stream causes small fluctuation in the value represented by the bit stream. This desirable error tolerance feature is extremely attractive for NML technology. By participating in this project, students will learn fascinating properties of nanomagnets, become proficient with micromagnetic simulation tools, simulate different stochastic NML circuit structures, and investigate the performance and power of these structures. Bolder students will get a chance to try out their own stochastic NML circuit designs.

Pre-requisite courses or skills: Basic knowledge in digital logic, computer architecture and electronic circuits are required. Knowledge in advanced circuit theory, probability/statistics and prior research experience in related areas are desired. Project: Exploiting the Power of GPU as Hardware Accelerators Faculty name and email: Dr. Sharon Hu, shu@nd.edu ND College: College of Engineering Project description: Graphic Processing Unit (GPU) has become an effective platform for accelerating computationally expensive applications. Many GPU-based systems have been developed for applications from a wide range of domains. However, simply porting a sequential program to GPU cannot really harness the raw compute power of GPU. Techniques in program parallelization, memory partitioning and sharing are needed. Our group has developed GPU-based solutions for a number of medical applications and some key functions in finite element analysis. The student intern will have an opportunity to work with graduate students to evaluate different design options in order to improve the performance of existing implementations and develop new implementations. Pre-requisite courses or skills: Knowledge of computer architecture and basic GPU programming is required. Familiarity with parallel algorithm design is desired. Project: Probabilistic impact analysis Faculty name and email: Dr. Raul Santelices, rsanteli@nd.edu ND College: College of Engineering Project description: Programmers are constantly making changes to software but often overlook the consequences of those changes. Change-impact analysis comes to the rescue by providing a picture of what parts of the software may be affected by a change. However, current techniques provide coarse impact results and often mark the whole software as impacted, which is not very useful. To solve this problem, this project will develop algorithms that compute measures of the strength of the impact of a change on different parts of the system. This way, developers can focus their attention on the most affected parts first. (Alternatively, the impact strength of a change on component A can be seen as the probability that the change affects the execution of A.) Pre-requisite courses or skills: Strong analytical and programming skills; good knowledge of Java

Project: Visualization of changes and dependences in software Faculty name and email: Dr. Raul Santelices, rsanteli@nd.edu ND College: College of Engineering Project description: For programmers to understand the role of a code fragment in a software system (how the fragment affects the rest of the system) and to decide what changes they can make in those fragments without unexpected side effects, they need to visualize the potential impacts of the fragment in a system. This project aims at creating an interactive visualization of the strengths of dependencies and change impacts in which the system is represented by a 2-dimensional map where colors are used to show the degree of impact of a code fragment on each point in the system. Developers can zoom in for more detail and zoom out for a bigger picture. Ideally, this visualization will be a plug-in for Eclipse where developers can switch between the code editors and the visualization via clicks or menu options. Pre-requisite courses or skills: Strong analytical and programming skills; good knowledge of Java      Project: Functional approximation of program fragments Faculty name and email: Dr. Raul Santelices, rsanteli@nd.edu ND College: College of Engineering Project description: To enable automatic test generation and program-behavior prediction, it is imperative to model a program or fragments of it by abstracting some features away while retaining essential aspects of their behavior. One such model proposed in this project is the approximation of a program by looking at it as a function where the input is the initial program state and the output is the state at the end. If a function can give a good approximation of what the end state is in terms of the initial state, it can be used to manipulate the input in order to force the program to produce specific outputs, such as errors that later need debugging. The approximating function can be created using machine learning, algebraic techniques, or a combination of multiple approaches. Pre-requisite courses or skills: Strong analytical and programming skills; good knowledge of Java      Project: Continuous analysis and testing while coding

Faculty name and email: Dr. Raul Santelices, rsanteli@nd.edu ND College: College of Engineering Project description: Modern integrated development environments (IDEs) such as Eclipse and Visual Studio have made the life of the programmer much easier. The IDE can quickly provide a wealth of information about a project and the relationships among its components or individual lines of code. However, IDEs work in a reactive way. The next generation of IDEs must be proactive, and this project proposes one technology that will speed the creation and maintenance of reliable code. We propose a task running in the background (and exploiting idle parallel resources) that executes existing or new analysis algorithms that produce new tests and find errors in the fragment of code currently under construction. As the programmer modifies the fragment, some tests become obsolete but others remain valid. The surviving tests will reveal errors detected during coding that the programmer missed after completing the coding task. Pre-requisite courses or skills: Strong analytical and programming skills; good knowledge of Java     Project: Characterization and Modeling of GaN-based Transistors for Millimeter-wave Circuits Faculty name and email: Dr. Patrick Fay, pfay@nd.edu ND College: College of Engineering Project description: GaN-based devices are emerging not only for power applications, but for high-speed, high-performance applications as well. In this project, agressively-scaled GaN-based devices will be characterized (DC, CV, RF/millimeter-wave) and models developed to describe the observed performance. Physical insights into the internal operation of the devices (e.g. dispersive phenomena) are of particular interest, and the use of the characterization to more fully understand these effects will be a key focus of the project. Work will include on-wafer testing of devices (DC measurements, low- and high-frequency CV, and on-wafer RF-millimeterwave (10 MHz-220 GHz) network analysis), as well as development of models suitable for integration in computer-aided design software packages. Pre-requisite courses or skills: Students with familiarity and/or interest in RF/microwave measurement techniques and hands-on experimental work are encouraged to apply.      Project: Nanosensors for Micro-RNA Profiling – Designing Medical Devices for Personalized Medicine

Faculty name and email: Dr. Hsueh-Chia Chang, hchang@nd.edu ND College: College of Engineering Project description: We are designing nanooptical sensors to allow enumeration of single molecules for a large numer of target molecules. One particular application is profiling (quantification) of a panel of regulatory miRNAs for cancer and chronic disease screening. The main technical challenges for such nanosensors are to amplify the optical signals significantly, so that quantification resolution is down to a single molecule, with simple LED light sources and miniature cameras. Such sensitivity enhancement is provided by nanostructures with conic geometries. Conic nanopipettes fabricated with laser-assisted drawing amplifies electric fields by 5 orders of magnitude to assemble a small number of nanoparticles for plasmonic amplification. Optical fiber bundles are wet-etched into nanocone arrays, with each 10-nm cone tip having greatly enhanced scattering and surface energy to attract binding nanoparticles and target molecules. We also design conformation-switching hairpin FRET oligo-probes for the nanoparticles so the entire molecular assay is label free. Such low-cost nanofabrication technologies are integrated to manufacture a medical platform for miRNA profiling that can soon be used in clinics and, at a later stage, at home for personalized disease screening. (Figure Caption: conic optical fiber array with 3-micron base and 10 nm tip).

Pre-requisite courses or skills: We are seeking highly motivated and extremely creative BS and MS students to work with a dynamic group of interdisciplinary and international researchers. Experience with microfluidics, plasmonics, biosensing or nanofabrication is preferred but not necessary. Project: Critical meta-path discovery in heterogeneous information networks Faculty name and email: Dr. Tim Weninger, tweninge@nd.edu

ND College: College of Engineering Project description: Similarity search is the primary function of databases and Web search engines. Recent work in large-scale information networks that consist of multi-typed, interconnected objects, such as the Web and social networks, we are able to study similarity these heterogeneous information networks. Intuitively, two objects are similar if they are linked by many paths in the network. However, existing similarity measures are defined for untyped/homogeneous information networks. Different semantic meanings behind paths are not taken into consideration. By considering different linkage paths in a network, one could derive various similarity semantics. {\em Meta path-based similarity} was introduced to find paths consisting of a sequence of relations defined between different object types (i.e., structural paths at the type or meta-level). However, meta path-based similarity requires handcrafted type-queries, only operates under strict conditions, and fails to scale to even moderately sized data sets. We propose to develop a framework that mines critical meta paths from large-scale heterogeneous information networks. Results of this proposed line of work would be able to not only describe if two or more objects are similar, but also how they are similar. Pre-requisite courses or skills: None. Two or three students desired for the project.