APS Logo

Discovering novel electrolytes by leveraging high-throughput experimentation and machine learning

ORAL

Abstract

Electrolytes play a critical role in governing the performance of numerous electrochemical energy storage and generation devices. For example, secondary battery technologies, such as lithium and sodium metal batteries, require electrolytes which exhibit both high ionic conductivity and electrochemical stability against the metal anode to maintain adequate cycle life. Unfortunately, few electrolytes exhibit both high ionic conductivity and high electrochemical stability, reducing the cyclability of Li and Na metal batteries and limiting their commercial viability. Thus, the discovery of novel electrolytes for these electrochemical devices is critical to their efficient and safe operation, as well as their commercial development. Unfortunately, the design space for electrolyte materials is rather large, rendering the discovery of new electrolytes for specific applications a laborious and inefficient process. To accelerate this process, we leverage the use of high-throughput robotic formulation and characterization to efficiently screen large libraries of electrolytes for Li and Na metal battery applications. This high-throughput platform is then coupled with machine learning to identify promising new candidates for screening. In this talk, we will discuss the application of this approach to two systems: solid polymer electrolytes and multicomponent “high entropy” liquid electrolytes. First, we highlight the utility of our high-throughput characterization tools by demonstrating its ability to rapidly and accurately measure ionic conductivities of model poly(ethylene oxide) electrolyte systems. Second, we demonstrate how coupling high-throughput experimentation and machine learning can allow for rapid screening and identification of new multicomponent liquid electrolytes with promising transport properties. Our results suggest that coupling high-throughput electrolyte characterization and machine learning is a promising approach to the design and discovery of novel electrolytes for a wide array of electrochemical energy storage and generation applications.

Presenters

  • Everett S Zofchak

    Massachusetts Institute of Technology

Authors

  • Everett S Zofchak

    Massachusetts Institute of Technology

  • Vanesa Munoz Perales

    Massachusetts Institute of Technology

  • Jason K Phong

    Massachusetts Institute of Technology

  • Sawyer D Cawthern

    Massachusetts Institute of Technology

  • Christian Plaza-Rivera

    Massachusetts Institute of Technology

  • Jeremiah A Johnson

    Massachusetts Institute of Technology

  • Yang Shao-Horn

    Massachusetts Institute of Technology