Michael Stolberg1,Yang Shao-Horn1,Jeremiah Johnson1
Massachusetts Institute of Technology1
Michael Stolberg1,Yang Shao-Horn1,Jeremiah Johnson1
Massachusetts Institute of Technology1
Polymeric electrolytes are a potential electrolyte material in the push towards all solid-state lithium-ion batteries, replacing the highly flammable liquids of the modern day. However, since the 1990s, the maximum ionic conductivity they can achieve has largely stagnated at approximately 10<sup>-3</sup> S/cm even at elevated temperatures, an order of magnitude below liquid electrolytes. Additionally, the highest performing polymer electrolytes are still based on polyethylene oxide (or other polyethers) with dissolved lithium salts. Our group seeks to modernize research into polymer electrolytes by taking a high throughput data driven approach which combines active machine learning with a high throughput instrument designed specifically to accelerate the pace and reliability of sample formulation and characterization. In this way we hope to further understand descriptors of ion conduction in polymeric materials, and ultimately increase the ionic conductivity of polymer electrolytes. Herein we will present our custom high throughput instrument and a growing database of polymer electrolytes formulated and characterized on the tool. Additionally, we will detail our active learning workflow and progress towards incorporation of multiple data streams such as molecular dynamics simulations, machine learned predictions, and experimental data with the common goal of boosting ionic conductivity in polymer electrolytes.