Balancing performance and sustainability in new materials discovery via machine learning

Background

Many of the most exciting new developments in engineering and technology can be traced back to the discovery of new materials which exhibit exciting new properties. However, one of the main challenges in discovering new materials is that the process has traditionally been a high-risk, low-reward endeavor with low odds of identifying a new material with exciting new properties. In fact, many of the most exciting new materials from high temperature superconductors to teflon fluoropolymers have been discovered by "accident" by fortuity and chance. Moreover, even when materials are designed for a given application little thought is given to economic, environmental, and other sustainability factors that may limit the eventual deployment of new materials. For example, rare-earth laden magnets, noble metals in catalysts, or toxic heavy-metals in solar cells and batteries. Going forward, a fundamentally new approach must be utilized that will enable the discovery of new materials at a radically increased pace by reducing the risk associated with exploring chemical "whitespace." One exciting new field, which the PI is a pioneer in, is using machine learning techniques to predict material performance before synthesis ever occurs. The advent of big data in materials science and other science fields has put machine learning in a position to provide exciting new developments that far outpace traditional one-at-a-time synthetic approaches. However, at the heart of machine learning is gathering, curating, and preparing data to be utilized as training datasets in machine learning algorithms. There is an enormous amount of work to be done in collecting this data especially with an eye towards sustainability metrics.

Student Role

The student involved with this project will work hand-in-hand with two graduate students who are funded under a recent NSF CAREER award on the discovery of new thermoelectric materials using machine learning, data mining, and materials informatics. The student will help gather new data from literature, organize and edit existing data, and write code to perform data science and predictions for new materials. Once a list of outstanding candidates is identified the student will help actually synthesize these materials via synthetic techniques like arc-melting, SPS sintering, precipitation, solid-state sintering, combustion synthesis, and microwave synthesis. Moreover, students will need to characterize the crystal structure, microstructure, thermal and electrical properties for materials that we synthesize in the lab.

Outcomes

The student will learn broad skills in coding, data science, and experimental aspects. Data science is a rapidly growing field and the student will learn how to employ machine learning techniques such as random forest algorithms, support vector machines, neural networks, principal component analysis, and regressions as well as leave-one-out-validation, descriptor and algorithm development. With regards to experiments, the student will learn state-of-the-art metal and ceramic material synthesis as well as microscopy, x-ray diffraction, thermal and electrical transport measurements. The student will learn how to work on a team and will become an expert in the field of thermoelectrics, which are renewable energy devices that convert waste heat to electricity.

Taylor Sparks
Assistant Professor

Materials Science & Engineering
College of Engineering
Global Change and Sustainability Center

The student will join the Sparks Research Group which currently has 5 PhD students, 2 MS students, 3 postdocs, and ~10 undergraduate researchers. We have a vibrant and exciting research endeavor that encompasses many areas of materials science related to energy, fuel conversion, and biomaterials with funding from the Department of Defense, Department of Energy, National Science Foundation, and private industries. For example, the student will attend weekly group meetings where we discuss Na-ion battery cathodes, Na-ion electrolytes, Na-ion material processing, kinetics of electrochemical devices, microbial coal to natural gas biogasification, bioceramics for percutaneous implants, CO2-derived carbon nanomaterials, synthetic diamond manufacture, fluoropolymers for waste-water treatment facilities, superhard materials discovery, electrochromic devices and more! I will make sure to meet with the individual student as well as their research team weekly. I will also aid in poster/figure development, writing academic articles, and preparing the student to present at a professional society conference in their field.